I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
1
6
,
No
.
4
,
A
u
g
u
s
t
20
2
6
,
p
p
.
1
985
~
1
9
9
7
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
1
6
i
4
.
pp
1
9
8
5
-
1
9
9
7
1985
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
SS
-
AN
FIS:
a
sem
i
-
superv
ised n
eur
o
-
fuzzy mo
del
for
offline
sig
na
ture
v
erifica
tion
Sa
dly
Sy
a
m
s
ud
din
,
J
ufr
i,
Su
ci
Ra
hm
a
Da
ni Ra
chm
a
n,
S
ury
a
ni,
Wilem
M
us
u,
Sa
lm
ia
t
i,
Yesy
cha
Arun Ma
ng
o
po
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
m
a
t
i
c
s,
D
i
p
a
U
n
i
v
e
r
si
t
y
M
a
k
a
ss
a
r
,
M
a
k
a
ss
a
r
,
I
n
d
o
n
e
s
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
2
4
,
2
0
2
5
R
ev
is
ed
Ma
y
1
4
,
2
0
2
6
Acc
ep
ted
J
u
l 2
2
,
2
0
2
6
S
ig
n
a
t
u
re
v
e
rifi
c
a
ti
o
n
re
m
a
in
s
a
c
rit
ica
l
a
u
t
h
e
n
ti
c
a
ti
o
n
m
e
c
h
a
n
ism
i
n
a
c
a
d
e
m
ic
a
n
d
a
d
m
in
istrati
v
e
e
n
v
ir
o
n
m
e
n
ts,
y
e
t
m
a
n
u
a
l
v
e
rifi
c
a
ti
o
n
is
v
u
l
n
e
ra
b
le
t
o
fo
r
g
e
ry
a
n
d
su
b
je
c
ti
v
e
j
u
d
g
m
e
n
t.
Th
is
stu
d
y
p
ro
p
o
se
s
S
S
-
AN
F
IS
,
a
se
m
i
-
su
p
e
rv
ise
d
n
e
u
ro
-
f
u
z
z
y
m
o
d
e
l
fo
r
o
ffli
n
e
sig
n
a
tu
re
v
e
rifi
c
a
ti
o
n
u
n
d
e
r
li
m
it
e
d
lab
e
l
e
d
d
a
ta
c
o
n
d
it
io
n
s.
T
h
e
p
ro
p
o
s
e
d
m
o
d
e
l
in
teg
ra
tes
p
se
u
d
o
-
la
b
e
l
-
b
a
se
d
s
e
lf
-
train
in
g
in
to
a
Ta
k
a
g
i
-
S
u
g
e
n
o
-
Ka
n
g
a
d
a
p
ti
v
e
n
e
u
ro
-
f
u
z
z
y
in
fe
re
n
c
e
sy
ste
m
(AN
F
IS
).
S
tatic
ima
g
e
-
b
a
se
d
fe
a
tu
re
s
we
re
e
x
trac
ted
fro
m
o
ffli
n
e
s
ig
n
a
t
u
re
ima
g
e
s
a
n
d
tra
n
sfo
rm
e
d
u
sin
g
p
rin
c
i
p
a
l
c
o
m
p
o
n
e
n
t
a
n
a
ly
sis
(P
C
A)
b
e
fo
re
c
las
sifica
ti
o
n
.
Ex
p
e
rim
e
n
ts
we
re
c
o
n
d
u
c
ted
o
n
8
0
0
o
ffli
n
e
si
g
n
a
tu
re
sa
m
p
les
c
o
ll
e
c
ted
fro
m
Dip
a
Un
iv
e
rsity
M
a
k
a
ss
a
r,
c
o
n
sistin
g
o
f
4
0
0
g
e
n
u
in
e
a
n
d
4
0
0
f
o
rg
e
d
sig
n
a
t
u
re
s.
Th
e
p
ro
p
o
se
d
m
o
d
e
l
a
c
h
iev
e
d
a
n
a
c
c
u
ra
c
y
o
f
9
0
.
5
%
,
p
re
c
isio
n
o
f
9
8
.
8
%
,
re
c
a
ll
o
f
8
2
.
0
%
,
a
n
d
F
1
-
sc
o
re
o
f
9
0
.
0
%
.
Th
e
h
ig
h
p
re
c
isio
n
i
n
d
i
c
a
tes
th
a
t
SS
-
AN
F
IS
is
e
ffe
c
ti
v
e
i
n
m
in
i
m
izin
g
fa
lse
p
o
siti
v
e
p
re
d
ictio
n
s,
wh
ich
is
imp
o
rtan
t
fo
r
a
c
a
d
e
m
ic
d
o
c
u
m
e
n
t
v
e
rifi
c
a
ti
o
n
.
T
h
e
re
su
lt
s
sh
o
w
th
a
t
th
e
p
ro
p
o
se
d
m
o
d
e
l
p
ro
v
id
e
s
a
p
ra
c
ti
c
a
l,
in
terp
re
tab
le,
a
n
d
c
o
m
p
u
tatio
n
a
ll
y
e
fficie
n
t
a
p
p
ro
a
c
h
fo
r
o
ffli
n
e
sig
n
a
tu
re
v
e
rifi
c
a
ti
o
n
,
p
a
rti
c
u
larly
i
n
in
stit
u
ti
o
n
a
l
se
tt
in
g
s wit
h
li
m
it
e
d
lab
e
led
d
a
ta.
K
ey
w
o
r
d
s
:
Ad
ap
tiv
e
n
e
u
r
o
-
f
u
zz
y
in
f
er
e
n
ce
s
y
s
tem
Fo
r
g
er
y
d
etec
tio
n
Prin
cip
al
co
m
p
o
n
e
n
t
an
al
y
s
is
,
d
ig
ital secu
r
ity
Sem
i
-
s
u
p
er
v
is
ed
lear
n
in
g
Sig
n
atu
r
e
r
ec
o
g
n
itio
n
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Sad
ly
Sy
am
s
u
d
d
in
Dep
ar
tm
en
t o
f
I
n
f
o
r
m
atics,
Dip
a
Un
iv
er
s
ity
Ma
k
ass
ar
J
l.
Per
in
tis
Kem
er
d
ek
aa
n
No
.
KM
.
9
,
T
am
alan
r
ea
I
n
d
ah
,
T
a
m
alan
r
ea
Dis
tr
ict,
Ma
k
ass
ar
C
ity
,
So
u
th
Su
lawe
s
i 9
0
2
4
5
,
I
n
d
o
n
esia
E
m
ail: sad
ly
s
@
u
n
d
ip
a.
ac
.
id
1.
I
NT
RO
D
UCT
I
O
N
Sig
n
atu
r
e
-
b
ased
a
u
th
en
ticatio
n
r
em
ain
s
wid
ely
u
s
ed
f
o
r
v
alid
atin
g
o
f
f
icial
d
o
cu
m
en
ts
in
ac
ad
em
ic,
ad
m
in
is
tr
ativ
e,
f
in
an
cial,
a
n
d
leg
al
en
v
ir
o
n
m
en
ts
b
ec
au
s
e
o
f
its
s
im
p
licity
,
cu
ltu
r
al
ac
ce
p
tan
ce
,
an
d
ea
s
e
o
f
im
p
lem
en
tatio
n
[
1
]
,
[
2
]
.
H
o
wev
er
,
m
an
u
al
s
ig
n
atu
r
e
v
er
if
icatio
n
is
v
u
ln
er
a
b
le
to
f
o
r
g
er
y
an
d
m
a
y
co
m
p
r
o
m
is
e
d
o
cu
m
e
n
t
au
th
en
ticity
,
in
s
titu
tio
n
al
cr
ed
ib
i
lity
,
an
d
leg
al
ac
co
u
n
tab
ilit
y
,
p
ar
ticu
lar
ly
in
ac
ad
em
ic
d
o
c
u
m
en
ts
s
u
ch
a
s
th
esis
ap
p
r
o
v
al
s
h
ee
ts
,
ex
am
in
atio
n
f
o
r
m
s
,
r
ec
o
m
m
en
d
atio
n
letter
s
,
an
d
ce
r
tific
ates
[
3
]
,
[
4
]
.
At
Dip
a
Un
iv
er
s
ity
Ma
k
ass
ar
,
s
ig
n
atu
r
es
ar
e
r
o
u
tin
ely
u
s
ed
to
v
alid
ate
ac
ad
em
ic
an
d
ad
m
in
is
tr
ativ
e
d
o
c
u
m
en
ts
is
s
u
ed
b
y
lectu
r
er
s
a
n
d
s
taf
f
.
R
ec
en
t
s
u
s
p
ec
ted
f
o
r
g
er
y
ca
s
es
in
d
icate
th
at
m
an
u
al
v
er
if
icatio
n
b
ased
o
n
ly
o
n
v
is
u
al
in
s
p
ec
tio
n
is
p
r
o
n
e
to
in
co
n
s
is
ten
cy
,
f
atig
u
e,
an
d
h
u
m
a
n
er
r
o
r
.
T
h
e
r
ef
o
r
e
,
a
n
au
to
m
ated
a
n
d
r
eliab
le
s
ig
n
at
u
r
e
v
e
r
if
icatio
n
s
y
s
tem
is
n
ee
d
ed
to
s
u
p
p
o
r
t
in
s
titu
tio
n
al
d
o
cu
m
en
t
s
ec
u
r
ity
an
d
r
ed
u
ce
s
u
b
jectiv
ity
.
Sig
n
atu
r
e
v
e
r
if
icatio
n
is
g
en
er
ally
d
iv
id
e
d
in
t
o
o
n
lin
e
a
n
d
o
f
f
lin
e
a
p
p
r
o
ac
h
es.
On
lin
e
s
ig
n
atu
r
e
v
er
if
icatio
n
u
s
es
d
y
n
am
ic
in
f
o
r
m
atio
n
c
o
llected
d
u
r
in
g
s
ig
n
in
g
,
s
u
c
h
as
p
en
p
r
ess
u
r
e,
w
r
itin
g
s
p
ee
d
,
s
tr
o
k
e
o
r
d
er
,
an
d
p
en
tr
ajec
to
r
y
,
wh
er
ea
s
o
f
f
lin
e
v
er
if
icatio
n
r
elie
s
o
n
ly
o
n
s
tatic
s
ig
n
atu
r
e
im
ag
es
[
5
]
,
[
6
]
.
Sin
ce
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
4
,
Au
g
u
s
t
20
2
6
:
1
9
8
5
-
1
997
1986
d
y
n
am
ic
in
f
o
r
m
atio
n
is
u
n
av
ailab
le
in
o
f
f
lin
e
im
ag
es,
o
f
f
lin
e
s
ig
n
atu
r
e
v
er
if
icatio
n
is
m
o
r
e
ch
allen
g
in
g
b
ec
au
s
e
th
e
s
y
s
tem
m
u
s
t
d
i
s
tin
g
u
is
h
g
en
u
in
e
an
d
f
o
r
g
e
d
s
ig
n
atu
r
es
u
s
in
g
o
n
ly
v
is
u
al
an
d
s
tr
u
ctu
r
al
ch
ar
ac
ter
is
tics
[
6
]
,
[
7
]
.
Var
io
u
s
co
m
p
u
tatio
n
al
m
eth
o
d
s
h
av
e
b
ee
n
p
r
o
p
o
s
ed
f
o
r
s
ig
n
atu
r
e
v
e
r
if
icatio
n
.
T
r
a
d
itio
n
a
l
m
ac
h
in
e
lear
n
in
g
m
eth
o
d
s
s
u
ch
as
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
i
n
e
(
SVM)
,
lo
g
is
tic
r
eg
r
ess
io
n
,
an
d
ar
tific
ial
n
eu
r
al
n
etwo
r
k
class
if
ier
s
ca
n
ac
h
iev
e
ac
ce
p
tab
le
p
er
f
o
r
m
a
n
ce
wh
e
n
f
e
atu
r
e
r
ep
r
esen
tatio
n
is
well
d
esig
n
ed
[
8
]
,
[
9
]
.
Ho
wev
er
,
th
ese
m
eth
o
d
s
o
f
ten
s
tr
u
g
g
le
with
u
n
ce
r
tain
t
y
,
h
i
g
h
in
tr
a
-
class
v
ar
iatio
n
,
an
d
o
v
er
lap
p
in
g
p
atter
n
s
b
etwe
en
g
en
u
in
e
an
d
f
o
r
g
e
d
s
ig
n
atu
r
es.
Dee
p
lear
n
i
n
g
m
et
h
o
d
s
,
p
a
r
ticu
lar
ly
c
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NN)
,
ca
n
au
to
m
atica
lly
lea
r
n
d
is
cr
im
in
ativ
e
v
is
u
al
f
ea
tu
r
e
s
f
r
o
m
s
ig
n
atu
r
e
im
ag
es
[
1
0
]
,
[
1
1
]
.
Nev
er
th
eless
,
C
NN
-
b
ased
m
o
d
els g
en
er
ally
r
eq
u
ir
e
lar
g
e
lab
eled
d
atasets
an
d
h
ig
h
co
m
p
u
tatio
n
al
r
eso
u
r
ce
s
,
wh
ich
m
ay
n
o
t
b
e
p
r
ac
tical
f
o
r
s
m
all
-
s
ca
le
ac
ad
em
ic
en
v
ir
o
n
m
en
ts
.
T
h
e
ad
ap
tiv
e
n
eu
r
o
-
f
u
zz
y
in
f
er
en
ce
s
y
s
tem
(
ANFI
S)
o
f
f
er
s
an
alter
n
ativ
e
ap
p
r
o
ac
h
b
y
co
m
b
in
in
g
n
eu
r
al
lear
n
in
g
ca
p
ab
ilit
y
wi
th
f
u
zz
y
r
ea
s
o
n
in
g
[
1
2
]
.
T
h
i
s
s
tr
u
ctu
r
e
is
s
u
itab
le
f
o
r
m
o
d
elin
g
n
o
n
lin
ea
r
,
u
n
ce
r
tain
,
an
d
am
b
ig
u
o
u
s
p
att
er
n
s
,
m
ak
in
g
ANFI
S
r
elev
an
t
f
o
r
o
f
f
lin
e
s
ig
n
atu
r
e
v
e
r
if
icatio
n
[
1
3
]
,
[
1
4
]
,
[
1
5
]
.
Ho
wev
er
,
co
n
v
e
n
tio
n
al
ANFI
S
s
till
d
ep
en
d
s
o
n
s
u
f
f
icien
t
lab
eled
d
ata
to
b
u
ild
r
eliab
le
f
u
zz
y
r
u
les
an
d
m
em
b
er
s
h
ip
f
u
n
ctio
n
s
.
W
h
en
lab
eled
s
am
p
les
ar
e
lim
ited
,
its
g
en
er
aliza
tio
n
ca
p
ab
ilit
y
m
a
y
d
ec
r
ea
s
e.
Sem
i
-
s
u
p
er
v
is
ed
lear
n
in
g
ad
d
r
ess
es
th
is
lim
itatio
n
b
y
u
tili
zin
g
b
o
th
lab
eled
an
d
u
n
lab
eled
d
ata
d
u
r
in
g
tr
ain
in
g
[
1
6
]
,
[
1
7
]
.
T
h
r
o
u
g
h
m
ec
h
an
is
m
s
s
u
ch
as
p
s
eu
d
o
-
lab
elin
g
,
co
n
f
i
d
en
t
p
r
e
d
ictio
n
s
f
r
o
m
u
n
la
b
eled
s
am
p
les
ca
n
b
e
in
co
r
p
o
r
ated
i
n
to
t
h
e
tr
ain
i
n
g
p
r
o
ce
s
s
to
im
p
r
o
v
e
g
en
er
aliza
tio
n
with
o
u
t
r
eq
u
ir
in
g
ex
ten
s
iv
e
m
an
u
a
l
an
n
o
tatio
n
[
1
8
]
.
Alth
o
u
g
h
s
em
i
-
s
u
p
er
v
is
ed
lear
n
in
g
h
as b
ee
n
wid
ely
e
x
p
lo
r
e
d
in
m
ac
h
in
e
lear
n
in
g
an
d
d
ee
p
lear
n
in
g
,
its
in
teg
r
atio
n
with
ANFI
S
f
o
r
o
f
f
lin
e
s
ig
n
atu
r
e
v
er
if
icatio
n
r
em
ain
s
lim
ited
.
M
o
s
t
ex
is
tin
g
s
tu
d
ies
f
o
cu
s
o
n
f
u
lly
s
u
p
er
v
is
ed
ANFI
S
m
o
d
els,
co
n
v
en
tio
n
al
class
if
ier
s
,
o
r
C
NN
-
b
ased
m
eth
o
d
s
.
Few
s
tu
d
ies
h
av
e
ex
am
in
ed
h
o
w
p
s
eu
d
o
-
lab
elin
g
ca
n
b
e
em
b
ed
d
e
d
in
to
th
e
ANFI
S
tr
ain
in
g
p
r
o
ce
s
s
to
im
p
r
o
v
e
p
e
r
f
o
r
m
an
ce
u
n
d
er
lim
ited
lab
ele
d
d
ata
co
n
d
itio
n
s
.
I
n
ad
d
itio
n
,
o
f
f
lin
e
s
ig
n
atu
r
e
v
e
r
if
icatio
n
r
eq
u
ir
es
s
tatic
im
ag
e
-
b
ased
f
ea
tu
r
e
ex
tr
ac
tio
n
,
s
u
ch
as
g
eo
m
etr
ic,
tex
tu
r
e,
an
d
g
r
ad
ien
t
-
b
ased
d
escr
ip
to
r
s
[
1
9
]
,
[
2
0
]
.
Prin
cip
al
co
m
p
o
n
e
n
t
an
al
y
s
is
(
PC
A)
ca
n
also
b
e
u
s
ed
to
r
ed
u
ce
f
ea
tu
r
e
d
im
e
n
s
io
n
ality
wh
ile
p
r
eser
v
in
g
in
f
o
r
m
ativ
e
v
ar
ian
ce
[
2
1
]
.
B
ased
o
n
th
ese
co
n
s
id
er
atio
n
s
,
th
is
s
tu
d
y
p
r
o
p
o
s
es
a
s
em
i
-
s
u
p
er
v
is
ed
ad
ap
tiv
e
n
eu
r
o
-
f
u
zz
y
in
f
er
en
ce
s
y
s
te
m
(
SS
-
AN
FIS)
f
o
r
o
f
f
lin
e
s
ig
n
atu
r
e
v
er
if
icatio
n
in
th
e
ac
ad
em
ic
en
v
ir
o
n
m
en
t
o
f
Dip
a
Un
iv
er
s
ity
Ma
k
ass
ar
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
in
teg
r
ates
p
s
eu
d
o
-
lab
elin
g
in
to
th
e
ANFI
S
tr
ain
in
g
p
r
o
ce
s
s
to
u
tili
ze
b
o
th
lab
eled
an
d
u
n
lab
eled
s
ig
n
atu
r
e
d
ata.
Un
lik
e
a
p
p
r
o
ac
h
es
th
at
r
ely
o
n
d
y
n
am
ic
s
ig
n
in
g
f
ea
tu
r
es,
th
is
s
tu
d
y
f
o
cu
s
es
ex
clu
s
iv
ely
o
n
s
tatic
im
ag
e
-
b
ased
f
ea
tu
r
es,
m
a
k
in
g
it
co
n
s
is
ten
t
with
o
f
f
lin
e
s
ig
n
atu
r
e
v
er
if
icatio
n
s
ce
n
ar
io
s
.
T
h
e
m
ai
n
co
n
tr
ib
u
t
io
n
s
o
f
t
h
is
s
tu
d
y
ar
e:
i
)
d
e
v
e
lo
p
in
g
a
n
SS
-
ANFI
S
f
r
am
ew
o
r
k
b
ased
o
n
s
em
i
-
s
u
p
er
v
is
ed
p
s
eu
d
o
-
lab
elin
g
;
ii)
u
s
in
g
s
tatic
im
ag
e
-
b
ased
f
ea
tu
r
es
s
u
p
p
o
r
te
d
b
y
g
e
o
m
etr
ic,
tex
tu
r
e,
g
r
ad
ie
n
t
-
b
ased
,
an
d
PC
A
-
r
ed
u
ce
d
r
ep
r
esen
tatio
n
s
;
iii)
ap
p
ly
in
g
a
f
air
ex
p
er
im
en
tal
p
r
o
to
co
l
in
wh
ich
all
b
aselin
e
m
o
d
els
ar
e
ev
alu
ated
u
s
in
g
th
e
s
am
e
d
ataset,
f
ea
tu
r
e
ex
tr
a
ctio
n
p
r
o
ce
s
s
,
an
d
m
et
r
ics;
an
d
iv
)
d
e
v
elo
p
in
g
a
Stre
am
lit
-
b
ased
p
r
o
to
ty
p
e
f
o
r
p
r
ac
tical
ac
ad
em
ic
d
o
cu
m
en
t
au
th
en
ticatio
n
.
T
o
f
u
r
th
e
r
p
o
s
itio
n
th
is
s
tu
d
y
am
o
n
g
e
x
is
tin
g
s
ig
n
atu
r
e
v
e
r
if
icatio
n
ap
p
r
o
ac
h
es,
T
ab
le
1
i
n
Ap
p
en
d
ix
s
u
m
m
ar
izes th
e
s
tr
en
g
th
s
an
d
lim
itatio
n
s
o
f
tr
ad
itio
n
al
m
ac
h
in
e
lear
n
in
g
,
d
e
ep
lear
n
in
g
,
co
n
v
en
tio
n
al
ANFI
S,
an
d
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
m
o
d
el.
T
r
a
d
itio
n
al
class
if
ier
s
ar
e
s
im
p
le
an
d
co
m
p
u
tatio
n
all
y
ef
f
icien
t
b
u
t
o
f
ten
s
tr
u
g
g
le
with
n
o
n
lin
ea
r
an
d
o
v
er
la
p
p
in
g
s
ig
n
atu
r
e
p
atte
r
n
s
.
C
NN
-
b
ased
m
eth
o
d
s
p
r
o
v
id
e
s
tr
o
n
g
im
a
g
e
r
ec
o
g
n
itio
n
ca
p
a
b
ilit
y
b
u
t
r
e
q
u
ir
e
lar
g
e
lab
eled
d
atasets
an
d
h
ig
h
er
co
m
p
u
tatio
n
al
r
eso
u
r
ce
s
.
C
o
n
v
en
tio
n
al
ANFI
S
o
f
f
er
s
in
ter
p
r
eta
b
ilit
y
an
d
is
s
u
itab
le
f
o
r
u
n
ce
r
tain
h
an
d
wr
itten
s
ig
n
atu
r
e
p
atter
n
s
,
b
u
t
its
p
er
f
o
r
m
an
ce
m
ay
d
ec
lin
e
wh
en
lab
eled
d
a
ta
ar
e
lim
ited
.
T
h
er
ef
o
r
e,
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
m
o
d
el
p
r
o
v
id
es
a
p
r
ac
tical
alter
n
ativ
e
f
o
r
o
f
f
lin
e
s
ig
n
atu
r
e
v
er
if
icatio
n
in
ac
ad
em
ic
en
v
ir
o
n
m
en
ts
with
lim
ited
lab
eled
s
am
p
les.
2.
M
E
T
H
O
D
T
h
is
r
esear
ch
em
p
lo
y
e
d
an
e
x
p
er
im
en
tal
q
u
an
titativ
e
ap
p
r
o
ac
h
to
d
e
v
elo
p
an
d
ev
alu
ate
a
h
y
b
r
id
m
o
d
el
f
o
r
f
o
r
g
ed
s
ig
n
atu
r
e
d
etec
tio
n
.
T
h
e
m
eth
o
d
o
lo
g
y
co
n
s
is
ted
o
f
s
ev
er
al
s
tag
es:
d
ataset
ac
q
u
is
itio
n
,
p
r
ep
r
o
ce
s
s
in
g
,
f
ea
tu
r
e
ex
t
r
ac
tio
n
,
m
o
d
el
d
ev
elo
p
m
en
t
u
s
in
g
ANFI
S
en
h
an
ce
d
with
s
em
i
-
s
u
p
er
v
is
ed
lear
n
in
g
,
an
d
ev
alu
atio
n
o
f
m
o
d
el
p
er
f
o
r
m
an
ce
.
T
h
e
wo
r
k
f
lo
w
is
illu
s
tr
ated
in
Fig
u
r
e
1
.
2
.
1
.
Da
t
a
s
et
a
cquis
it
io
n
T
h
e
d
ataset
was
co
llected
f
r
o
m
lectu
r
er
s
an
d
s
taf
f
o
f
Dip
a
Un
iv
er
s
ity
Ma
k
ass
ar
with
f
o
r
m
al
co
n
s
en
t.
A
to
tal
o
f
8
0
0
o
f
f
lin
e
s
ig
n
at
u
r
e
s
am
p
les
wer
e
o
b
tain
ed
,
co
n
s
is
tin
g
o
f
4
0
0
g
en
u
in
e
s
ig
n
atu
r
es
an
d
4
0
0
f
o
r
g
e
d
s
ig
n
atu
r
es.
T
h
e
g
en
u
i
n
e
s
ig
n
a
tu
r
es
wer
e
co
llected
d
ir
ec
tly
f
r
o
m
th
e
co
r
r
esp
o
n
d
i
n
g
lectu
r
e
r
s
an
d
s
taf
f
,
wh
ile
th
e
f
o
r
g
ed
s
ig
n
atu
r
es we
r
e
p
r
e
p
ar
ed
to
s
im
u
late
d
o
cu
m
e
n
t v
e
r
if
icatio
n
s
ce
n
ar
io
s
in
ac
a
d
em
i
c
ad
m
in
is
tr
atio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
SS
-
A
N
F
I
S
:
a
s
emi
-
s
u
p
ervis
ed
n
eu
r
o
-
fu
z
z
y
mo
d
el
fo
r
o
ffli
n
e
s
ig
n
a
tu
r
e
ve
r
ifica
tio
n
(
S
a
d
ly
S
ya
msu
d
d
in
)
1987
Fig
u
r
e
1
.
R
esear
ch
wo
r
k
f
lo
w
o
f
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S m
o
d
el
T
h
e
d
ataset
was
o
r
g
an
ized
in
t
o
two
lab
eled
ca
teg
o
r
ies:
g
en
u
in
e
an
d
f
o
r
g
ed
.
Ge
n
u
in
e
s
ig
n
a
tu
r
es
wer
e
en
co
d
ed
as
class
0
,
wh
ile
f
o
r
g
ed
s
ig
n
atu
r
es
wer
e
en
c
o
d
ed
as
class
1
.
T
h
e
d
ataset
was
u
s
ed
f
o
r
m
o
d
el
t
r
ain
in
g
,
v
alid
atio
n
,
an
d
ev
alu
atio
n
.
T
h
is
d
ataset
d
esig
n
r
ef
lects
a
r
ea
lis
tic
ac
ad
em
ic
d
o
cu
m
en
t
v
er
if
icatio
n
s
ettin
g
,
wh
er
e
th
e
s
y
s
tem
m
u
s
t
d
is
tin
g
u
is
h
au
th
e
n
tic
s
ig
n
atu
r
es
f
r
o
m
v
is
u
ally
s
im
ilar
f
o
r
g
ed
s
ig
n
at
u
r
es.
Alth
o
u
g
h
th
e
d
ataset
is
lim
ited
to
an
in
s
tit
u
tio
n
al
s
ettin
g
,
it
p
r
o
v
i
d
es
a
b
alan
ce
d
ex
p
er
im
en
tal
b
asis
f
o
r
ev
alu
atin
g
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S m
o
d
el
in
an
ac
ad
em
ic
d
o
cu
m
e
n
t v
er
if
ic
atio
n
s
ce
n
ar
io
.
2
.
2
.
P
re
pro
ce
s
s
ing
R
aw
s
ig
n
atu
r
e
im
ag
es we
r
e
p
r
ep
r
o
ce
s
s
ed
to
en
s
u
r
e
co
n
s
is
ten
cy
an
d
r
ed
u
ce
n
o
is
e
:
a.
Gr
ay
s
ca
le
co
n
v
er
s
io
n
to
elim
i
n
ate
co
lo
r
v
ar
iatio
n
s
b.
R
esizin
g
to
a
s
tan
d
ar
d
ized
r
es
o
lu
tio
n
o
f
3
0
0
d
p
i f
o
r
u
n
if
o
r
m
ity
c.
No
is
e
r
ed
u
ctio
n
to
r
em
o
v
e
b
ac
k
g
r
o
u
n
d
a
r
tifa
cts an
d
p
e
n
s
m
u
d
g
es
d.
No
r
m
aliza
tio
n
o
f
p
ix
el
in
ten
s
i
ty
v
alu
es
T
h
is
s
tep
en
s
u
r
ed
th
at
th
e
i
n
p
u
t d
ata
wer
e
u
n
if
o
r
m
a
n
d
o
p
tim
ized
f
o
r
f
ea
tu
r
e
e
x
tr
ac
tio
n
.
2
.
3
.
F
e
a
t
ure
ex
t
r
a
ct
io
n
Sin
ce
th
is
s
tu
d
y
f
o
cu
s
es
o
n
o
f
f
lin
e
s
ig
n
atu
r
e
v
er
if
icatio
n
,
o
n
ly
s
tatic
im
ag
e
-
b
ased
f
ea
t
u
r
es
wer
e
ex
tr
ac
ted
f
r
o
m
s
ig
n
atu
r
e
im
a
g
es.
Dy
n
am
ic
f
ea
tu
r
es
s
u
ch
as
wr
itin
g
s
p
ee
d
,
p
en
p
r
ess
u
r
e,
s
tr
o
k
e
o
r
d
er
,
a
n
d
p
e
n
tr
ajec
to
r
y
wer
e
n
o
t u
s
ed
b
ec
au
s
e
th
ese
f
ea
tu
r
es a
r
e
n
o
t a
v
ailab
le
in
o
f
f
lin
e
s
ig
n
atu
r
e
im
a
g
es.
T
h
e
f
ea
tu
r
e
ex
tr
ac
tio
n
p
r
o
c
ess
f
o
cu
s
ed
o
n
v
is
u
al
an
d
s
tr
u
ctu
r
al
ch
ar
ac
ter
is
tics
o
f
s
ig
n
atu
r
es,
in
clu
d
in
g
g
eo
m
etr
ic
f
ea
tu
r
es,
tex
tu
r
e
in
f
o
r
m
atio
n
,
g
r
ad
ien
t
-
b
ased
d
escr
ip
to
r
s
,
an
d
d
im
e
n
s
io
n
ality
r
ed
u
ctio
n
u
s
in
g
PC
A
.
Geo
m
etr
ic
f
ea
tu
r
e
s
wer
e
ex
tr
ac
ted
to
r
ep
r
esen
t
th
e
s
h
ap
e,
wid
th
,
h
eig
h
t,
ar
ea
,
an
d
asp
ec
t
r
atio
o
f
th
e
s
ig
n
atu
r
e.
T
ex
t
u
r
e
-
b
ased
f
ea
tu
r
es
wer
e
u
s
ed
to
ca
p
tu
r
e
g
r
ay
s
ca
le
in
ten
s
ity
v
ar
iatio
n
s
,
wh
ile
g
r
ad
ien
t
-
b
ased
d
escr
ip
to
r
s
wer
e
u
s
ed
to
r
ep
r
esen
t
ed
g
e
an
d
s
tr
o
k
e
o
r
ien
tatio
n
p
atter
n
s
.
PC
A
w
as
th
en
ap
p
lied
to
r
ed
u
ce
f
ea
tu
r
e
d
im
en
s
io
n
ality
an
d
r
etain
th
e
m
o
s
t
in
f
o
r
m
ati
v
e
co
m
p
o
n
e
n
ts
b
ef
o
r
e
class
if
icatio
n
.
T
h
is
f
ea
t
u
r
e
ex
tr
ac
tio
n
s
tr
ateg
y
en
s
u
r
es
t
h
at
th
e
p
r
o
p
o
s
ed
m
o
d
el
is
co
n
s
is
ten
t
with
th
e
n
atu
r
e
o
f
o
f
f
lin
e
s
ig
n
atu
r
e
v
er
if
icatio
n
,
wh
e
r
e
th
e
in
p
u
t d
ata
co
n
s
is
t o
n
ly
o
f
s
tatic
s
ig
n
a
tu
r
e
im
ag
es.
2
.
4
.
SS
-
ANF
I
S
m
o
del dev
elo
pm
ent
T
h
e
ANFI
S
was
im
p
lem
en
ted
to
class
if
y
g
en
u
in
e
an
d
f
o
r
g
ed
s
ig
n
atu
r
es.
ANFI
S
co
m
b
in
es
th
e
r
ea
s
o
n
in
g
ca
p
ab
ilit
y
o
f
f
u
zz
y
in
f
er
en
ce
s
y
s
tem
s
with
th
e
l
ea
r
n
in
g
ca
p
ab
ilit
y
o
f
n
e
u
r
al
n
etwo
r
k
s
,
m
ak
in
g
it
s
u
itab
le
f
o
r
u
n
ce
r
tain
an
d
am
b
ig
u
o
u
s
d
ata
s
u
ch
as
h
an
d
wr
itten
s
ig
n
atu
r
es
[
1
2
]
,
[
2
4
]
,
[
2
5
]
.
T
o
ad
d
r
ess
th
e
lim
itatio
n
o
f
lab
eled
d
ata,
a
s
em
i
-
s
u
p
er
v
is
ed
lear
n
in
g
s
tr
a
teg
y
was
in
co
r
p
o
r
ate
d
b
ased
o
n
p
s
eu
d
o
-
lab
elin
g
p
r
in
cip
les
[
1
6
]
,
[
1
7
]
,
[
1
8
]
.
T
h
is
allo
wed
th
e
m
o
d
el
to
lev
e
r
ag
e
b
o
t
h
lab
eled
an
d
u
n
lab
el
ed
d
ata,
im
p
r
o
v
in
g
g
en
er
aliza
tio
n
a
n
d
r
o
b
u
s
tn
ess
in
r
ea
l
-
wo
r
ld
ac
ad
em
ic
s
ce
n
ar
io
s
.
Similar
h
y
b
r
id
ap
p
r
o
ac
h
es
co
m
b
i
n
in
g
ANFI
S
with
o
th
er
tech
n
iq
u
e
s
h
av
e
also
d
em
o
n
s
tr
ated
s
t
r
o
n
g
p
er
f
o
r
m
an
ce
in
r
elate
d
d
o
m
ain
s
,
s
u
ch
as
in
tr
u
s
io
n
d
etec
tio
n
[
1
3
]
an
d
m
o
b
ile
m
alwa
r
e
p
r
e
d
ictio
n
[
1
4
]
.
T
h
e
ar
ch
itectu
r
e
o
f
th
e
ANF
I
S
m
o
d
el
is
s
h
o
wn
in
Fig
u
r
e
2
.
T
h
e
ANFI
S
ar
ch
itectu
r
e
co
n
s
is
ts
o
f
s
ev
er
al
s
eq
u
en
tial
lay
er
s
.
T
h
e
in
p
u
t
lay
e
r
r
ec
eiv
es
PC
A
-
tr
a
n
s
f
o
r
m
ed
s
tatic
im
ag
e
f
ea
tu
r
e
s
.
T
h
e
f
u
zz
if
icatio
n
lay
er
m
ap
s
th
e
c
r
is
p
in
p
u
t
v
a
lu
es
in
to
Gau
s
s
ian
m
em
b
er
s
h
ip
f
u
n
ctio
n
s
.
T
h
e
r
u
le
lay
er
c
o
m
p
u
tes
th
e
f
ir
in
g
s
tr
en
g
th
o
f
ea
ch
f
u
zz
y
r
u
le,
wh
ile
th
e
n
o
r
m
aliza
tio
n
lay
e
r
s
ca
les th
e
f
ir
in
g
s
tr
en
g
th
s
.
T
h
e
d
ef
u
zz
if
icatio
n
lay
e
r
ap
p
lies
lin
ea
r
Su
g
en
o
co
n
s
eq
u
en
ts
,
an
d
th
e
o
u
tp
u
t
lay
er
p
r
o
d
u
ce
s
th
e
f
in
al
b
in
ar
y
d
ec
is
io
n
,
n
am
ely
g
en
u
in
e
o
r
f
o
r
g
ed
.
Fo
r
clar
ity
,
th
e
f
u
ll
p
r
o
ce
s
s
i
s
s
u
m
m
ar
ized
as
a
f
lo
wc
h
ar
t
in
Fig
u
r
e
2
.
T
h
e
d
iag
r
a
m
illu
s
tr
ates
th
e
lin
ea
r
f
lo
w
f
r
o
m
in
p
u
t
ac
q
u
i
s
itio
n
th
r
o
u
g
h
p
r
e
p
r
o
ce
s
s
in
g
,
f
ea
tu
r
e
ex
tr
ac
tio
n
,
ANFI
S
class
if
icatio
n
,
s
em
i
-
s
u
p
er
v
is
ed
en
h
a
n
ce
m
en
t,
e
v
al
u
atio
n
,
an
d
f
in
al
o
u
tp
u
t
v
ia
th
e
Stre
am
lit ap
p
licatio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
4
,
Au
g
u
s
t
20
2
6
:
1
9
8
5
-
1
997
1988
St
art
In
p
u
t
(Si
g
n
at
u
re
Ima
g
es
)
Prep
ro
ces
s
i
n
g
(G
ray
s
c
al
e
,
Re
s
i
ze,
N
o
i
s
e
Rem
o
v
al
)
Feat
u
re
E
x
t
ra
ct
i
o
n
(PC
A
,
St
at
i
c
&
d
y
n
am
i
c
feat
u
res
)
E
v
a
l
u
at
i
o
n
(A
ccu
racy
,
Pr
ec
i
s
i
o
n
,
R
ec
al
l
,
F1
-
Sc
o
re
)
O
u
t
p
u
t
(G
en
u
i
n
e/
F
o
rg
ed
,
S
t
re
aml
i
t
U
I)
A
N
F
IS
M
o
d
el
(Fu
zzy
+
N
eu
ral
N
et
w
o
rk
)
Sem
i
-
s
u
p
er
v
i
s
ed
E
n
h
an
ce
m
en
t
(U
s
e
s
u
n
l
ab
e
l
ed
d
at
a)
Fi
n
i
s
h
Fig
u
r
e
2
.
Flo
wch
ar
t
o
f
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S m
o
d
el
2
.
5
.
P
r
o
po
s
ed
SS
-
ANF
I
S
a
lg
o
rit
hm
T
h
e
p
r
o
p
o
s
ed
Sem
i
-
Su
p
er
v
i
s
ed
Ad
ap
tiv
e
Neu
r
o
-
Fu
zz
y
I
n
f
er
e
n
ce
Sy
s
tem
(
SS
-
ANFI
S)
was
im
p
lem
en
ted
u
s
in
g
a
s
elf
-
tr
ain
in
g
s
tr
ateg
y
with
p
s
eu
d
o
-
lab
el
in
g
.
T
h
e
tr
ain
in
g
p
r
o
ce
d
u
r
e
co
n
s
is
ts
o
f
two
m
ain
s
tag
es:
s
u
p
er
v
is
ed
in
itializati
o
n
an
d
s
em
i
-
s
u
p
er
v
is
ed
r
ef
in
em
en
t.
I
n
th
e
s
u
p
e
r
v
is
ed
in
iti
aliza
tio
n
s
tag
e,
th
e
lab
eled
s
ig
n
atu
r
e
d
ataset
was
d
iv
id
ed
in
to
tr
ain
in
g
a
n
d
v
alid
atio
n
s
u
b
s
ets
u
s
in
g
s
tr
atif
ied
s
p
litt
in
g
.
T
h
e
ex
tr
ac
ted
s
tatic
im
ag
e
-
b
ased
f
ea
tu
r
es
wer
e
s
tan
d
ar
d
ized
u
s
in
g
Stan
d
ar
d
Scaler
a
n
d
t
h
en
tr
an
s
f
o
r
m
e
d
u
s
in
g
PC
A
.
I
n
th
is
im
p
lem
en
tatio
n
,
PC
A
was
co
n
f
ig
u
r
ed
with
s
i
x
co
m
p
o
n
en
ts
an
d
wh
iten
in
g
en
ab
led
t
o
o
b
tain
co
m
p
ac
t a
n
d
d
ec
o
r
r
elate
d
f
ea
t
u
r
e
r
ep
r
esen
tatio
n
s
.
T
h
e
tr
an
s
f
o
r
m
ed
f
ea
t
u
r
es
wer
e
u
s
ed
as
in
p
u
ts
to
a
T
ak
a
g
i
-
Su
g
en
o
-
Kan
g
ANFI
S
clas
s
if
ier
.
T
h
e
ANFI
S
m
o
d
el
u
s
ed
Gau
s
s
ian
m
em
b
er
s
h
ip
f
u
n
ctio
n
s
in
t
h
e
an
tece
d
en
t
lay
er
a
n
d
lin
ea
r
Su
g
en
o
c
o
n
s
eq
u
e
n
ts
f
o
r
b
in
ar
y
class
if
icatio
n
.
T
h
e
m
o
d
el
was
tr
ain
e
d
u
s
in
g
th
e
A
d
am
o
p
tim
izer
a
n
d
b
in
ar
y
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
with
lo
g
its
.
Af
ter
s
u
p
er
v
is
ed
tr
ain
in
g
,
th
e
s
em
i
-
s
u
p
er
v
is
ed
s
tag
e
was
ap
p
lied
wh
en
u
n
lab
eled
s
a
m
p
les
wer
e
av
ailab
le.
T
h
e
tr
ain
ed
ANFI
S
m
o
d
el
p
r
ed
icted
t
h
e
p
r
o
b
a
b
ilit
y
o
f
ea
c
h
u
n
lab
eled
s
am
p
le
b
elo
n
g
in
g
to
th
e
f
o
r
g
ed
class
.
Sam
p
les
with
h
ig
h
p
r
e
d
ictio
n
co
n
f
id
en
c
e
wer
e
s
elec
ted
f
o
r
p
s
eu
d
o
-
lab
elin
g
.
I
n
th
is
im
p
lem
en
tatio
n
,
s
am
p
les with
p
r
o
b
ab
ilit
y
g
r
ea
ter
th
a
n
o
r
eq
u
al
to
0
.
8
wer
e
ass
ig
n
ed
to
th
e
f
o
r
g
ed
class
,
wh
ile
s
am
p
les
with
p
r
o
b
ab
ilit
y
les
s
th
an
o
r
eq
u
al
t
o
0
.
2
wer
e
ass
ig
n
ed
to
th
e
g
en
u
in
e
cl
ass
.
Sam
p
les
with
p
r
o
b
a
b
ilit
ies b
etwe
en
th
ese
th
r
esh
o
ld
s
wer
e
ex
clu
d
e
d
to
r
e
d
u
ce
th
e
r
is
k
o
f
i
n
co
r
r
ec
t
p
s
eu
d
o
-
lab
el
ass
ig
n
m
en
t.
T
o
av
o
id
class
im
b
alan
ce
d
u
r
i
n
g
p
s
eu
d
o
-
lab
el
ex
p
a
n
s
io
n
,
th
e
n
u
m
b
er
o
f
p
s
eu
d
o
-
lab
eled
s
a
m
p
les wa
s
lim
ited
to
a
m
ax
im
u
m
o
f
5
0
s
am
p
les
p
er
class
in
ea
ch
s
elf
-
tr
ain
in
g
r
o
u
n
d
.
T
h
e
s
elec
ted
p
s
eu
d
o
-
lab
eled
s
am
p
les
wer
e
th
en
co
m
b
in
e
d
with
th
e
o
r
ig
in
al
lab
eled
d
ata
s
et,
an
d
th
e
ANFI
S
m
o
d
el
w
as
f
in
e
-
tu
n
ed
u
s
in
g
th
e
ex
p
an
d
e
d
tr
ain
in
g
s
et.
T
h
is
s
em
i
-
s
u
p
er
v
is
ed
r
ef
in
em
en
t
was
r
ep
ea
ted
f
o
r
two
r
o
u
n
d
s
o
r
s
to
p
p
ed
ea
r
lier
if
n
o
h
ig
h
-
co
n
f
id
en
ce
u
n
lab
ele
d
s
am
p
les we
r
e
av
ailab
le.
T
h
is
p
r
o
ce
d
u
r
e
allo
ws
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
m
o
d
el
to
ex
p
lo
it
u
n
lab
eled
s
ig
n
at
u
r
e
s
am
p
les
wh
ile
co
n
tr
o
llin
g
p
s
eu
d
o
-
la
b
el
n
o
is
e
th
r
o
u
g
h
co
n
f
id
en
ce
-
b
ased
s
am
p
le
s
elec
tio
n
an
d
class
-
b
alan
ce
d
p
s
eu
d
o
-
lab
el
in
clu
s
io
n
.
Alg
o
r
ith
m
1
.
SS
-
ANFI
S tr
ain
in
g
p
r
o
ce
d
u
r
e
b
ased
o
n
p
s
eu
d
o
-
lab
el
s
elf
-
tr
ain
in
g
I
n
p
u
t
:
1.
La
b
e
l
e
d
f
e
a
t
u
r
e
se
t
D
l
=
{(
x
i
,
y
i
)
}
2.
U
n
l
a
b
e
l
e
d
f
e
a
t
u
r
e
se
t
D
u
=
{
x
j
}
3.
P
C
A
d
i
me
n
s
i
o
n
d
=
6
4.
N
u
mb
e
r
o
f
f
u
z
z
y
r
u
l
e
s
R
=
6
5.
S
u
p
e
r
v
i
se
d
l
e
a
r
n
i
n
g
r
a
t
e
=
1
e
-
3
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
SS
-
A
N
F
I
S
:
a
s
emi
-
s
u
p
ervis
ed
n
eu
r
o
-
fu
z
z
y
mo
d
el
fo
r
o
ffli
n
e
s
ig
n
a
tu
r
e
ve
r
ifica
tio
n
(
S
a
d
ly
S
ya
msu
d
d
in
)
1989
6.
F
i
n
e
-
t
u
n
i
n
g
l
e
a
r
n
i
n
g
r
a
t
e
=
5
e
-
4
7.
N
u
mb
e
r
o
f
se
l
f
-
t
r
a
i
n
i
n
g
r
o
u
n
d
s =
2
8.
F
a
k
e
c
o
n
f
i
d
e
n
c
e
t
h
r
e
s
h
o
l
d
=
0
.
8
9.
R
e
a
l
c
o
n
f
i
d
e
n
c
e
t
h
r
e
s
h
o
l
d
=
0
.
2
1
0
.
M
a
x
i
m
u
m
p
s
e
u
d
o
-
l
a
b
e
l
e
d
s
a
m
p
l
e
s p
e
r
c
l
a
s
s =
5
0
O
u
t
p
u
t
:
Tr
a
i
n
e
d
S
S
-
A
N
F
I
S
mo
d
e
l
P
r
o
c
e
d
u
r
e
:
1.
S
p
l
i
t
t
h
e
l
a
b
e
l
e
d
d
a
t
a
se
t
D
l
i
n
t
o
t
r
a
i
n
i
n
g
a
n
d
v
a
l
i
d
a
t
i
o
n
s
u
b
se
t
s
u
si
n
g
s
t
r
a
t
i
f
i
e
d
sam
p
l
i
n
g
.
2.
S
t
a
n
d
a
r
d
i
z
e
t
h
e
e
x
t
r
a
c
t
e
d
f
e
a
t
u
r
e
v
e
c
t
o
r
s
u
si
n
g
S
t
a
n
d
a
r
d
S
c
a
l
e
r
.
3.
A
p
p
l
y
P
C
A
w
i
t
h
d
=
6
c
o
m
p
o
n
e
n
t
s a
n
d
w
h
i
t
e
n
i
n
g
t
o
r
e
d
u
c
e
f
e
a
t
u
r
e
d
i
m
e
n
si
o
n
a
l
i
t
y
.
4.
I
n
i
t
i
a
l
i
z
e
t
h
e
TSK
-
A
N
F
I
S
mo
d
e
l
w
i
t
h
R
=
6
f
u
z
z
y
r
u
l
e
s.
5.
Tr
a
i
n
t
h
e
i
n
i
t
i
a
l
A
N
F
I
S
mo
d
e
l
u
s
i
n
g
t
h
e
l
a
b
e
l
e
d
t
r
a
i
n
i
n
g
d
a
t
a
w
i
t
h
A
d
a
m
o
p
t
i
mi
z
e
r
a
n
d
b
i
n
a
r
y
c
r
o
ss
-
e
n
t
r
o
p
y
l
o
ss
w
i
t
h
l
o
g
i
t
s.
6.
Ev
a
l
u
a
t
e
t
h
e
i
n
i
t
i
a
l
su
p
e
r
v
i
se
d
A
N
F
I
S
mo
d
e
l
o
n
t
h
e
v
a
l
i
d
a
t
i
o
n
s
u
b
se
t
.
7.
P
r
e
d
i
c
t
t
h
e
f
o
r
g
e
d
-
c
l
a
ss
p
r
o
b
a
b
i
l
i
t
y
p
(
x
j
)
f
o
r
e
a
c
h
u
n
l
a
b
e
l
e
d
s
a
m
p
l
e
i
n
D
u
.
8.
S
e
l
e
c
t
h
i
g
h
-
c
o
n
f
i
d
e
n
c
e
f
o
r
g
e
d
sa
mp
l
e
s wh
e
r
e
p
(
x
j
)
≥
0
.
8
.
9.
S
e
l
e
c
t
h
i
g
h
-
c
o
n
f
i
d
e
n
c
e
g
e
n
u
i
n
e
sam
p
l
e
s wh
e
r
e
p
(
x
j
)
≤
0
.
2
.
1
0
.
A
ssi
g
n
p
se
u
d
o
-
l
a
b
e
l
1
t
o
s
e
l
e
c
t
e
d
f
o
r
g
e
d
s
a
m
p
l
e
s a
n
d
p
se
u
d
o
-
l
a
b
e
l
0
t
o
s
e
l
e
c
t
e
d
g
e
n
u
i
n
e
sam
p
l
e
s.
1
1
.
Li
mi
t
p
se
u
d
o
-
l
a
b
e
l
e
d
sa
mp
l
e
s
t
o
a
ma
x
i
m
u
m
o
f
5
0
sam
p
l
e
s
p
e
r
c
l
a
ss
i
n
e
a
c
h
se
l
f
-
t
r
a
i
n
i
n
g
r
o
u
n
d
.
1
2
.
C
o
m
b
i
n
e
t
h
e
o
r
i
g
i
n
a
l
l
a
b
e
l
e
d
d
a
t
a
set
a
n
d
s
e
l
e
c
t
e
d
p
s
e
u
d
o
-
l
a
b
e
l
e
d
sam
p
l
e
s.
1
3
.
R
e
f
i
t
S
t
a
n
d
a
r
d
S
c
a
l
e
r
a
n
d
P
C
A
u
s
i
n
g
t
h
e
e
x
p
a
n
d
e
d
t
r
a
i
n
i
n
g
d
a
t
a
s
e
t
.
1
4
.
F
i
n
e
-
t
u
n
e
t
h
e
A
N
F
I
S
mo
d
e
l
u
si
n
g
t
h
e
e
x
p
a
n
d
e
d
d
a
t
a
se
t
w
i
t
h
l
e
a
r
n
i
n
g
r
a
t
e
5
e
-
4.
1
5
.
R
e
p
e
a
t
S
t
e
p
s
7
–
1
4
f
o
r
t
w
o
se
l
f
-
t
r
a
i
n
i
n
g
r
o
u
n
d
s
o
r
st
o
p
e
a
r
l
i
e
r
i
f
n
o
h
i
g
h
-
c
o
n
f
i
d
e
n
c
e
u
n
l
a
b
e
l
e
d
s
a
m
p
l
e
s a
r
e
a
v
a
i
l
a
b
l
e
.
16.
R
e
t
u
r
n
t
h
e
f
i
n
a
l
S
S
-
A
N
F
I
S
p
i
p
e
l
i
n
e
c
o
n
s
i
st
i
n
g
o
f
S
t
a
n
d
a
r
d
S
c
a
l
e
r
,
P
C
A
,
a
n
d
t
r
a
i
n
e
d
TS
K
-
A
N
F
I
S
mo
d
e
l
.
T
h
e
f
in
al
im
p
lem
e
n
tatio
n
p
a
r
a
m
eter
s
o
f
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
m
o
d
el
ar
e
s
u
m
m
ar
ized
in
T
ab
le
2
.
T
h
ese
p
ar
am
eter
s
wer
e
s
elec
te
d
b
ased
o
n
th
e
im
p
lem
en
ted
tr
ain
in
g
co
n
f
i
g
u
r
atio
n
to
e
n
s
u
r
e
r
ep
r
o
d
u
cib
ilit
y
o
f
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S m
o
d
e
l.
T
ab
le
2
.
Hy
p
er
p
ar
a
m
eter
co
n
f
ig
u
r
atio
n
o
f
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S m
o
d
el
P
a
r
a
me
t
e
r
V
a
l
u
e
F
e
a
t
u
r
e
r
e
p
r
e
se
n
t
a
t
i
o
n
H
O
G
a
n
d
H
u
m
o
me
n
t
s
F
e
a
t
u
r
e
s
c
a
l
i
n
g
S
t
a
n
d
a
r
d
S
c
a
l
e
r
D
i
me
n
si
o
n
a
l
i
t
y
r
e
d
u
c
t
i
o
n
P
C
A
N
u
mb
e
r
o
f
P
C
A
c
o
m
p
o
n
e
n
t
s
6
P
C
A
w
h
i
t
e
n
i
n
g
Tr
u
e
P
C
A
r
a
n
d
o
m s
t
a
t
e
42
A
N
F
I
S
t
y
p
e
Ta
k
a
g
i
-
S
u
g
e
n
o
-
K
a
n
g
A
N
F
I
S
M
e
m
b
e
r
s
h
i
p
f
u
n
c
t
i
o
n
G
a
u
ss
i
a
n
N
u
mb
e
r
o
f
f
u
z
z
y
r
u
l
e
s
6
C
o
n
se
q
u
e
n
t
t
y
p
e
Li
n
e
a
r
S
u
g
e
n
o
A
c
t
i
v
a
t
i
o
n
f
u
n
c
t
i
o
n
S
i
g
m
o
i
d
Lo
ss f
u
n
c
t
i
o
n
B
i
n
a
r
y
c
r
o
ss
-
e
n
t
r
o
p
y
w
i
t
h
l
o
g
i
t
s
O
p
t
i
mi
z
e
r
A
d
a
m
S
u
p
e
r
v
i
se
d
l
e
a
r
n
i
n
g
r
a
t
e
1e
-
3
S
u
p
e
r
v
i
se
d
t
r
a
i
n
i
n
g
e
p
o
c
h
s
40
S
e
l
f
-
t
r
a
i
n
i
n
g
r
o
u
n
d
s
2
P
seu
d
o
-
l
a
b
e
l
t
h
r
e
s
h
o
l
d
f
o
r
f
o
r
g
e
d
c
l
a
s
s
p
≥
0
.
8
P
seu
d
o
-
l
a
b
e
l
t
h
r
e
s
h
o
l
d
f
o
r
g
e
n
u
i
n
e
c
l
a
ss
p
≤
0
.
2
M
a
x
i
m
u
m
p
s
e
u
d
o
-
l
a
b
e
l
e
d
s
a
m
p
l
e
s p
e
r
c
l
a
s
s
50
F
i
n
e
-
t
u
n
i
n
g
e
p
o
c
h
s
p
e
r
r
o
u
n
d
20
F
i
n
e
-
t
u
n
i
n
g
l
e
a
r
n
i
n
g
r
a
t
e
5e
-
4
D
e
c
i
s
i
o
n
t
h
r
e
sh
o
l
d
f
o
r
s
i
n
g
l
e
-
i
m
a
g
e
i
n
f
e
r
e
n
c
e
0
.
6
5
V
a
l
i
d
a
t
i
o
n
s
p
l
i
t
8
0
:
2
0
st
r
a
t
i
f
i
e
d
s
p
l
i
t
2
.
6
.
M
a
t
hema
t
ica
l
m
o
delin
g
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
co
m
b
in
es
P
C
A
f
o
r
f
ea
tu
r
e
d
im
en
s
io
n
ality
r
ed
u
ctio
n
an
d
th
e
ANFI
S
en
h
an
ce
d
with
s
em
i
-
s
u
p
er
v
is
ed
lear
n
in
g
to
im
p
r
o
v
e
class
if
icatio
n
p
er
f
o
r
m
an
ce
u
n
d
e
r
lim
ited
lab
eled
d
ata
co
n
d
itio
n
s
.
2
.
6
.
1
.
F
ea
t
ure
pro
j
ec
t
io
n
us
ing
P
CA
Giv
en
an
in
p
u
t
s
ig
n
atu
r
e
im
a
g
e
r
ep
r
esen
ted
as
a
r
aw
f
ea
tu
r
e
v
ec
to
r
∈
ℝ
d
.
PC
A
is
u
s
ed
to
r
e
d
u
ce
d
im
en
s
io
n
ality
:
=
(
−
)
(
1
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
4
,
Au
g
u
s
t
20
2
6
:
1
9
8
5
-
1
997
1990
wh
er
e
μ
is
th
e
m
ea
n
v
ec
to
r
,
an
d
W
co
n
tain
s
th
e
to
p
M
eig
en
v
ec
to
r
s
o
f
th
e
c
o
v
ar
ian
ce
m
atr
i
x
:
=
1
=
1
(
−
)
(
−
)
(
2
)
T
h
e
p
r
o
jecte
d
v
ec
to
r
z
∈
R
M
is
u
s
ed
as th
e
in
p
u
t to
ANFI
S
.
2
.
6
.
2
.
ANF
I
S
s
t
ruct
ure
E
ac
h
f
u
zz
y
r
u
le
r
is
f
o
r
m
u
lated
in
f
ir
s
t
-
o
r
d
er
Su
g
e
n
o
f
o
r
m
:
R
u
le
r:
IF
1
is
1
A
N
D
2
is
2
A
N
D
…
A
N
D
is
,
THEN
(
)
=
0
+
(
1
)
1
+
(
2
)
2
+
…
+
(
)
(
3
)
wh
er
e
r
ep
r
esen
ts
th
e
m
-
th
PC
A
-
tr
an
s
f
o
r
m
ed
in
p
u
t
f
ea
tu
r
e,
AM
r
d
en
o
tes th
e
f
u
zz
y
s
et
o
f
t
h
e
m
-
th
in
p
u
t in
r
u
le
,
ar
m
is
th
e
co
n
s
eq
u
e
n
t p
a
r
am
eter
,
an
d
(
)
is
th
e
lin
ea
r
Su
g
en
o
o
u
t
p
u
t o
f
r
u
le
.
Gau
s
s
ian
m
em
b
er
s
h
ip
f
u
n
ctio
n
s
ar
e
ad
o
p
te
d
f
o
r
f
u
zz
i
f
icatio
n
:
(
)
=
(
−
(
−
)
2
(
2
2
)
)
(
4
)
w
h
er
e
c
mr
an
d
σ
mr
d
en
o
te
th
e
c
en
ter
an
d
wid
t
h
p
ar
a
m
eter
s
T
h
e
f
ir
in
g
s
tr
en
g
th
o
f
r
u
le
r
is
:
=
(
)
(
5
)
an
d
n
o
r
m
alize
d
as:
̄
=
(
6
)
T
h
e
o
v
e
r
all
o
u
tp
u
t is:
=
̄
(
7
)
A
lo
g
is
tic
ac
tiv
atio
n
is
th
en
ap
p
lied
f
o
r
b
in
ar
y
class
if
icatio
n
(
r
ea
l v
s
.
f
ak
e
)
:
=
(
)
(
8
)
2
.
6
.
1
.
Sem
i
-
s
up
er
v
is
ed
enha
ncem
ent
T
h
e
s
em
i
-
s
u
p
er
v
is
ed
en
h
an
c
em
en
t
in
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
m
o
d
el
is
im
p
lem
en
ted
u
s
in
g
a
p
s
eu
d
o
-
lab
el
-
b
ased
s
elf
-
tr
ain
i
n
g
s
tr
ateg
y
.
Af
ter
th
e
in
itial
s
u
p
er
v
is
ed
ANFI
S
m
o
d
el
is
tr
ain
ed
u
s
in
g
lab
ele
d
s
ig
n
atu
r
e
s
am
p
les,
th
e
m
o
d
el
p
r
ed
icts
th
e
p
r
o
b
ab
ilit
y
o
f
ea
c
h
u
n
lab
ele
d
s
am
p
le
b
elo
n
g
in
g
to
th
e
f
o
r
g
ed
class
.
T
h
e
p
r
e
d
icted
p
r
o
b
ab
ilit
y
is
co
m
p
u
ted
as f
o
llo
ws:
(
)
=
(
(
)
)
(
9
)
wh
er
e
(
)
is
th
e
p
r
ed
icted
p
r
o
b
a
b
ilit
y
th
at
th
e
u
n
lab
eled
s
am
p
le
zu
b
elo
n
g
s
to
t
h
e
f
o
r
g
ed
cl
ass
,
σ
is
th
e
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ctio
n
,
a
n
d
(
)
is
th
e
o
u
tp
u
t lo
g
it o
f
th
e
ANFI
S m
o
d
el.
Ps
eu
d
o
-
lab
els
ar
e
ass
ig
n
ed
o
n
ly
to
u
n
lab
ele
d
s
am
p
les
wi
th
h
ig
h
p
r
e
d
ictio
n
c
o
n
f
id
e
n
c
e.
I
n
th
is
im
p
lem
en
tatio
n
,
a
n
u
n
lab
eled
s
am
p
le
is
ass
ig
n
ed
to
th
e
f
o
r
g
ed
class
if
(
)
≥
0
.
8
,
wh
ile
it
is
ass
ig
n
ed
to
th
e
g
en
u
in
e
class
if
(
)
≤
0
.
2
.
Sam
p
les
with
p
r
o
b
ab
ilit
y
v
alu
es
b
etwe
en
0
.
2
a
n
d
0
.
8
ar
e
e
x
cl
u
d
ed
f
r
o
m
th
e
p
s
eu
d
o
-
lab
elin
g
p
r
o
ce
s
s
to
r
e
d
u
ce
th
e
r
is
k
o
f
i
n
co
r
r
ec
t
p
s
e
u
d
o
-
lab
el
ass
ig
n
m
en
t.
T
h
e
p
s
eu
d
o
-
lab
elin
g
r
u
le
is
d
ef
in
ed
as f
o
llo
ws:
̂
=
1
,
if
(
)
≥
0
.
8
̂
=
0
,
if
(
)
≤
0
.
2
is
d
is
ca
r
d
ed
,
if
0
.
2
<
(
)
<
0
.
8
(
1
0
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
SS
-
A
N
F
I
S
:
a
s
emi
-
s
u
p
ervis
ed
n
eu
r
o
-
fu
z
z
y
mo
d
el
fo
r
o
ffli
n
e
s
ig
n
a
tu
r
e
ve
r
ifica
tio
n
(
S
a
d
ly
S
ya
msu
d
d
in
)
1991
w
h
er
e
̂
d
en
o
tes
th
e
p
s
eu
d
o
-
la
b
el
ass
ig
n
ed
to
th
e
u
n
lab
eled
s
am
p
le
.
T
h
e
lab
el
1
r
ep
r
ese
n
ts
th
e
f
o
r
g
e
d
class
,
wh
ile
th
e
lab
el
0
r
ep
r
ese
n
ts
th
e
g
en
u
in
e
class
.
T
o
r
ed
u
ce
class
im
b
alan
ce
d
u
r
in
g
p
s
eu
d
o
-
lab
el
e
x
p
an
s
io
n
,
th
e
n
u
m
b
er
o
f
s
elec
ted
p
s
eu
d
o
-
lab
ele
d
s
am
p
les
is
lim
ited
to
a
m
ax
i
m
u
m
o
f
5
0
s
am
p
les
p
er
class
in
ea
ch
s
elf
-
tr
ain
in
g
r
o
u
n
d
.
T
h
e
s
elec
ted
p
s
eu
d
o
-
lab
eled
s
am
p
les ar
e
th
en
c
o
m
b
in
ed
with
th
e
o
r
ig
in
al
lab
eled
d
ataset
to
f
o
r
m
a
n
ex
p
a
n
d
ed
tr
ain
in
g
s
et
:
=
∪
(
1
1
)
wh
er
e
Dl
d
en
o
tes th
e
o
r
ig
in
al
lab
eled
d
ataset
an
d
Dp
d
en
o
te
s
th
e
s
elec
ted
p
s
eu
d
o
-
lab
eled
d
ataset
.
T
h
e
SS
-
ANFI
S
m
o
d
el
is
th
en
f
in
e
-
tu
n
e
d
u
s
in
g
th
e
ex
p
a
n
d
ed
tr
ain
in
g
s
et.
T
h
e
tr
ain
in
g
o
b
jectiv
e
f
o
llo
ws b
in
ar
y
c
r
o
s
s
-
en
tr
o
p
y
with
lo
g
its
,
d
ef
in
ed
as
:
=
(
1
/
)
(
,
(
)
)
(
1
2
)
wh
er
e
is
th
e
tr
ain
in
g
lo
s
s
,
N
is
th
e
n
u
m
b
er
o
f
tr
ain
in
g
s
am
p
les
in
th
e
ex
p
an
d
ed
d
ataset,
y
i
is
th
e
tr
u
e
lab
el
o
r
p
s
eu
d
o
-
lab
el
o
f
s
am
p
le
,
(
)
is
th
e
o
u
tp
u
t
lo
g
it
o
f
th
e
ANFI
S
m
o
d
el,
a
n
d
B
C
E
d
en
o
tes
b
i
n
ar
y
c
r
o
s
s
-
en
tr
o
p
y
lo
s
s
.
T
h
is
s
em
i
-
s
u
p
er
v
is
ed
m
ec
h
an
is
m
en
ab
les
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
m
o
d
el
to
lear
n
f
r
o
m
b
o
th
m
an
u
ally
lab
eled
s
am
p
les
an
d
h
ig
h
-
co
n
f
i
d
en
ce
p
s
eu
d
o
-
l
ab
eled
s
am
p
les.
B
y
ex
clu
d
in
g
lo
w
-
co
n
f
id
en
ce
u
n
lab
eled
s
am
p
les an
d
lim
itin
g
th
e
n
u
m
b
er
o
f
p
s
eu
d
o
-
lab
ele
d
s
am
p
les p
er
class
,
th
e
m
o
d
el
r
ed
u
ce
s
th
e
r
is
k
o
f
n
o
is
y
p
s
eu
d
o
-
lab
els wh
ile
im
p
r
o
v
in
g
g
en
e
r
aliza
tio
n
u
n
d
er
li
m
ited
lab
eled
d
ata
c
o
n
d
itio
n
s
[
1
6
]
,
[
1
7
]
,
[
1
8
]
.
2
.
6
.
2
.
E
v
a
lua
t
io
n
m
e
t
rics
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
was
e
v
alu
ated
u
s
in
g
ac
cu
r
ac
y
,
p
r
e
cisi
o
n
,
r
ec
all,
an
d
F1
-
s
co
r
e.
Acc
u
r
ac
y
m
ea
s
u
r
es
th
e
p
r
o
p
o
r
tio
n
o
f
c
o
r
r
ec
tly
class
if
ied
s
ig
n
atu
r
es.
Pre
cisi
o
n
m
ea
s
u
r
es
th
e
p
r
o
p
o
r
tio
n
o
f
co
r
r
ec
tly
p
r
ed
icted
f
o
r
g
e
d
s
ig
n
atu
r
es
a
m
o
n
g
all
s
am
p
les
p
r
e
d
icted
a
s
f
o
r
g
ed
,
w
h
ile
r
ec
all
m
ea
s
u
r
es
th
e
ab
ilit
y
o
f
th
e
m
o
d
el
to
d
etec
t
ac
tu
al
f
o
r
g
ed
s
ig
n
atu
r
es.
T
h
e
F1
-
s
co
r
e
r
ep
r
e
s
en
ts
th
e
h
ar
m
o
n
ic
m
ea
n
o
f
p
r
ec
is
io
n
an
d
r
ec
all
,
p
r
o
v
id
i
n
g
a
b
alan
ce
d
e
v
alu
ati
o
n
o
f
class
if
icatio
n
r
eliab
ilit
y
an
d
d
etec
tio
n
ca
p
ab
ilit
y
.
T
h
e
tar
g
et
p
er
f
o
r
m
a
n
ce
was
s
et
at
a
m
in
im
u
m
o
f
9
0
%
ac
cu
r
ac
y
,
in
lin
e
with
p
r
ev
io
u
s
C
NN
-
b
ased
an
d
ANFI
S
-
b
ased
s
tu
d
ies
[
3
]
,
[
1
3
]
,
[
2
6
]
.
Ach
ie
v
in
g
t
h
is
th
r
e
s
h
o
ld
en
s
u
r
es
th
e
s
y
s
tem
’
s
co
m
p
etitiv
en
ess
with
s
tate
-
of
-
th
e
-
ar
t a
p
p
r
o
ac
h
es wh
ile
ad
d
r
ess
in
g
th
e
ch
allen
g
e
o
f
lim
ited
lab
eled
d
ata.
2
.
6
.
3
.
Su
m
m
a
r
y
o
f
m
e
t
ho
do
lo
g
y
T
h
e
o
v
e
r
all
m
eth
o
d
o
lo
g
ical
s
t
ag
es o
f
th
is
s
tu
d
y
ar
e
s
u
m
m
ar
i
ze
d
in
T
ab
le
3
.
T
ab
le
3
.
Su
m
m
a
r
y
o
f
m
eth
o
d
o
lo
g
y
S
t
a
g
e
D
e
t
a
i
l
s
D
a
t
a
s
e
t
A
c
q
u
i
si
t
i
o
n
8
0
0
s
a
m
p
l
e
s (4
0
0
g
e
n
u
i
n
e
s
i
g
n
a
t
u
r
e
s,
4
0
0
f
a
k
e
si
g
n
a
t
u
r
e
s)
f
r
o
m
D
i
p
a
U
n
i
v
e
r
si
t
y
M
a
k
a
ss
a
r
Le
c
t
u
r
e
r
s
P
r
e
p
r
o
c
e
ss
i
n
g
G
r
a
y
sc
a
l
e
c
o
n
v
e
r
si
o
n
,
r
e
s
i
z
e
(
3
0
0
d
p
i
)
,
n
o
i
s
e
r
e
m
o
v
a
l
,
n
o
r
mal
i
z
a
t
i
o
n
F
e
a
t
u
r
e
Ex
t
r
a
c
t
i
o
n
S
t
a
t
i
c
i
m
a
g
e
-
b
a
se
d
f
e
a
t
u
r
e
s:
g
e
o
me
t
r
i
c
f
e
a
t
u
r
e
s,
t
e
x
t
u
r
e
f
e
a
t
u
r
e
s,
g
r
a
d
i
e
n
t
-
b
a
se
d
d
e
scr
i
p
t
o
r
s
,
a
n
d
P
C
A
-
b
a
s
e
d
d
i
me
n
si
o
n
a
l
i
t
y
r
e
d
u
c
t
i
o
n
M
o
d
e
l
D
e
v
e
l
o
p
me
n
t
TSK
-
A
N
F
I
S
w
i
t
h
G
a
u
ss
i
a
n
m
e
m
b
e
r
s
h
i
p
f
u
n
c
t
i
o
n
s
a
n
d
l
i
n
e
a
r
S
u
g
e
n
o
c
o
n
se
q
u
e
n
t
s
S
e
mi
-
su
p
e
r
v
i
se
d
M
o
d
u
l
e
P
seu
d
o
-
l
a
b
e
l
-
b
a
se
d
se
l
f
-
t
r
a
i
n
i
n
g
u
s
i
n
g
h
i
g
h
-
c
o
n
f
i
d
e
n
c
e
sam
p
l
e
s
Ev
a
l
u
a
t
i
o
n
A
c
c
u
r
a
c
y
,
P
r
e
c
i
s
i
o
n
,
R
e
c
a
l
l
,
F
1
-
sc
o
r
e
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
E
x
perim
ent
a
l
s
et
up
T
h
e
ex
p
er
im
en
ts
wer
e
co
n
d
u
c
ted
u
s
in
g
a
d
ataset
o
f
h
a
n
d
wr
itten
s
ig
n
atu
r
es
c
o
llected
f
r
o
m
lectu
r
er
s
an
d
s
taf
f
at
Di
p
a
Un
i
v
e
rsity
M
a
k
a
ss
a
r
.
T
h
e
d
ataset
co
n
s
is
ted
o
f
a
to
tal
o
f
8
0
0
s
ig
n
at
u
r
e
s
a
m
p
les,
d
iv
id
ed
in
to
two
ca
teg
o
r
ies:
a.
R
ea
l sig
n
atu
r
es:
400
s
am
p
les
b.
Fo
r
g
ed
s
ig
n
atu
r
es:
400
s
am
p
le
s
T
o
en
s
u
r
e
r
eliab
le
e
v
alu
atio
n
,
th
e
d
ataset
was
s
p
lit
in
to
tr
ain
in
g
an
d
test
in
g
s
ets,
with
8
0
%
o
f
th
e
d
ata
u
s
ed
f
o
r
tr
ain
in
g
an
d
2
0
%
r
eser
v
ed
f
o
r
test
in
g
.
T
h
e
tr
ain
in
g
s
et
was
u
tili
ze
d
to
o
p
ti
m
ize
th
e
p
ar
am
eter
s
o
f
th
e
ANFI
S
,
wh
ile
th
e
test
i
n
g
s
et
was
em
p
lo
y
ed
to
e
v
alu
ate
th
e
g
en
er
aliza
tio
n
ca
p
ab
ilit
y
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el.
Pre
p
r
o
ce
s
s
in
g
s
tep
s
in
clu
d
e
d
g
r
ay
s
ca
le
co
n
v
er
s
io
n
,
r
esizin
g
to
3
0
0
d
p
i,
an
d
n
o
is
e
r
em
o
v
a
l.
Featu
r
e
ex
tr
ac
tio
n
was
p
er
f
o
r
m
e
d
u
s
in
g
s
tatic
im
ag
e
-
b
ased
f
ea
tu
r
es,
in
clu
d
in
g
g
eo
m
etr
ic
ch
ar
ac
ter
is
tics
,
tex
tu
r
e
in
f
o
r
m
atio
n
,
g
r
ad
ien
t
-
b
ased
d
escr
ip
to
r
s
,
an
d
PC
A
-
b
ased
d
i
m
en
s
io
n
ality
r
ed
u
ctio
n
.
C
o
n
s
is
ten
t
with
th
e
o
f
f
lin
e
s
ig
n
atu
r
e
v
er
if
icatio
n
s
ettin
g
,
n
o
d
y
n
am
ic
s
ig
n
in
g
f
ea
tu
r
es
wer
e
u
s
ed
in
th
e
e
x
p
er
im
e
n
t.
T
h
e
p
r
o
p
o
s
ed
SS
-
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
4
,
Au
g
u
s
t
20
2
6
:
1
9
8
5
-
1
997
1992
ANFI
S
m
o
d
el
was
tr
ain
ed
u
s
i
n
g
th
e
f
in
al
h
y
p
er
p
ar
am
eter
c
o
n
f
ig
u
r
atio
n
p
r
esen
ted
in
T
ab
le
2
,
in
clu
d
in
g
s
ix
PC
A
co
m
p
o
n
en
ts
,
s
ix
f
u
zz
y
r
u
les,
4
0
s
u
p
er
v
is
ed
tr
ai
n
in
g
e
p
o
ch
s
,
an
d
two
s
elf
-
tr
ain
i
n
g
r
o
u
n
d
s
.
3.
2
.
E
x
perim
ent
a
l R
esu
lt
s
T
h
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
m
o
d
el
was
ev
alu
ated
u
s
in
g
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
.
T
o
p
r
o
v
id
e
co
n
tex
tu
al
co
m
p
ar
is
o
n
,
T
ab
le
4
p
r
esen
ts
a
liter
atu
r
e
-
b
ased
co
m
p
ar
is
o
n
o
f
s
ev
er
a
l
s
ig
n
atu
r
e
v
er
if
icatio
n
m
eth
o
d
s
,
in
clu
d
in
g
L
o
g
is
tic
R
eg
r
ess
io
n
,
SVM,
C
NN,
co
n
v
en
tio
n
al
ANFI
S,
an
d
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
m
o
d
el.
I
t
s
h
o
u
ld
b
e
n
o
ted
th
at
th
e
r
ep
o
r
ted
b
aselin
e
r
esu
lts
m
ay
b
e
d
er
iv
e
d
f
r
o
m
d
if
f
er
en
t
s
tu
d
ies
an
d
m
a
y
i
n
v
o
lv
e
d
i
f
f
er
en
t
d
atasets
,
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
es,
an
d
ex
p
e
r
im
en
tal
p
r
o
to
co
ls
.
T
h
er
ef
o
r
e,
t
h
e
co
m
p
ar
is
o
n
is
in
ten
d
e
d
to
s
h
o
w
g
e
n
er
al
p
er
f
o
r
m
a
n
ce
tr
en
d
s
r
ath
e
r
th
an
a
d
ir
ec
t o
n
e
-
to
-
o
n
e
e
x
p
er
im
e
n
tal
b
en
ch
m
ar
k
.
T
ab
le
4
.
L
iter
atu
r
e
-
b
ased
c
o
m
p
ar
is
o
n
o
f
s
ig
n
atu
r
e
v
er
if
icatio
n
m
eth
o
d
s
M
e
t
h
o
d
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
[
9
]
7
8
.
2
%
7
6
.
5
%
7
7
.
1
%
7
6
.
8
%
S
V
M
[
8
]
,
[
2
2
]
8
4
.
6
%
8
3
.
2
%
8
4
.
0
%
8
3
.
6
%
C
N
N
(
b
a
s
e
l
i
n
e
)
[
1
0
]
,
[
1
1
]
,
[
2
3
]
9
0
.
3
%
8
9
.
7
%
9
0
.
1
%
8
9
.
9
%
A
N
F
I
S
[
1
2
]
,
[
2
4
]
8
8
.
0
%
8
7
.
5
%
8
7
.
9
%
8
7
.
7
%
P
r
o
p
o
se
d
S
S
-
A
N
F
I
S
9
0
.
5
%
9
8
.
8
%
8
2
.
0
%
9
0
.
0
%
As
s
h
o
wn
in
T
ab
le
4
,
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
ac
h
iev
ed
an
ac
cu
r
ac
y
o
f
9
0
.
5
%,
p
r
ec
is
io
n
o
f
9
8
.
8
%,
r
ec
all
o
f
8
2
.
0
%,
an
d
F1
-
s
co
r
e
o
f
9
0
.
0
%.
C
o
m
p
ar
ed
with
th
e
r
ep
o
r
ted
tr
e
n
d
s
in
th
e
liter
a
tu
r
e,
th
e
p
r
o
p
o
s
ed
m
o
d
el
s
h
o
ws
co
m
p
etitiv
e
ac
cu
r
ac
y
an
d
th
e
h
i
g
h
est
p
r
ec
is
io
n
am
o
n
g
th
e
co
m
p
ar
ed
m
eth
o
d
s
.
T
h
e
h
ig
h
p
r
ec
is
io
n
in
d
icate
s
th
at
th
e
m
o
d
el
p
r
o
d
u
ce
s
v
e
r
y
f
ew
f
alse
p
o
s
itiv
e
p
r
e
d
ictio
n
s
.
T
h
is
ch
ar
ac
ter
is
tic
is
p
ar
ticu
lar
ly
im
p
o
r
tan
t
in
ac
ad
em
ic
d
o
cu
m
en
t
v
e
r
if
icatio
n
,
w
h
er
e
in
co
r
r
ec
tly
class
if
y
in
g
g
e
n
u
in
e
s
ig
n
atu
r
es
as
f
o
r
g
ed
m
ay
lea
d
to
ad
m
in
is
tr
ativ
e
co
n
s
eq
u
e
n
ce
s
.
Alth
o
u
g
h
th
e
r
ec
all
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el
is
lo
wer
th
an
th
at
o
f
th
e
C
NN
b
aselin
e,
th
e
F1
-
s
co
r
e
r
em
ain
s
co
m
p
etitiv
e.
T
h
is
in
d
icate
s
th
at
th
e
p
r
o
p
o
s
ed
m
o
d
el
m
ain
tain
s
a
r
ea
s
o
n
ab
le
b
alan
ce
b
etwe
e
n
class
if
icatio
n
r
eliab
ilit
y
an
d
f
o
r
g
er
y
d
etec
tio
n
ca
p
a
b
ilit
y
.
T
h
e
r
esu
lt
also
s
u
g
g
ests
th
at
t
h
e
s
em
i
-
s
u
p
er
v
is
ed
en
h
an
ce
m
e
n
t
co
n
t
r
ib
u
tes
to
im
p
r
o
v
i
n
g
t
h
e
r
eliab
ilit
y
o
f
A
NFI
S
-
b
ased
class
if
icatio
n
u
n
d
er
lim
ited
la
b
eled
d
ata
co
n
d
itio
n
s
.
3.
3
.
Co
nfusi
o
n
m
a
t
rix
a
na
l
y
s
is
Fig
u
r
e
3
p
r
esen
ts
th
e
co
n
f
u
s
io
n
m
atr
ix
o
f
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
m
o
d
el.
T
h
e
m
o
d
e
l
co
r
r
ec
tly
class
if
ied
3
9
6
g
en
u
in
e
s
ig
n
at
u
r
es
an
d
3
2
8
f
o
r
g
ed
s
ig
n
atu
r
es.
On
ly
4
g
en
u
in
e
s
ig
n
atu
r
es
we
r
e
m
is
class
if
ied
a
s
f
o
r
g
ed
,
in
d
icatin
g
a
v
er
y
lo
w
f
alse
p
o
s
itiv
e
r
ate.
T
h
is
r
esu
lt
s
u
p
p
o
r
ts
th
e
h
ig
h
p
r
ec
is
io
n
v
alu
e
o
f
9
8
.
8
%
r
ep
o
r
ted
in
T
ab
le
4
.
Fig
u
r
e
3
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
m
o
d
e
l
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
SS
-
A
N
F
I
S
:
a
s
emi
-
s
u
p
ervis
ed
n
eu
r
o
-
fu
z
z
y
mo
d
el
fo
r
o
ffli
n
e
s
ig
n
a
tu
r
e
ve
r
ifica
tio
n
(
S
a
d
ly
S
ya
msu
d
d
in
)
1993
Ho
wev
er
,
7
2
f
o
r
g
e
d
s
ig
n
atu
r
es
wer
e
m
is
class
if
ied
as
g
en
u
in
e.
T
h
is
in
d
icate
s
th
at
s
o
m
e
f
o
r
g
ed
s
am
p
les
h
ad
v
is
u
al
ch
ar
ac
ter
is
tics
th
at
wer
e
h
ig
h
ly
s
im
ilar
to
g
en
u
in
e
s
ig
n
atu
r
es.
Su
ch
c
ases
ar
e
co
m
m
o
n
in
o
f
f
lin
e
s
ig
n
atu
r
e
v
e
r
if
icatio
n
,
p
ar
ticu
lar
ly
wh
en
f
o
r
g
e
d
s
ig
n
atu
r
es
clo
s
ely
im
itate
th
e
s
h
a
p
e,
s
tr
o
k
e
s
tr
u
ctu
r
e,
an
d
s
p
atial
d
is
tr
ib
u
tio
n
o
f
g
e
n
u
in
e
s
ig
n
atu
r
es.
T
h
er
e
f
o
r
e,
wh
ile
th
e
p
r
o
p
o
s
ed
m
o
d
el
is
h
ig
h
ly
r
eliab
le
in
m
in
im
izin
g
f
alse
ac
cu
s
atio
n
s
,
f
u
r
th
er
im
p
r
o
v
em
en
t
is
s
till
n
ee
d
ed
to
in
c
r
ea
s
e
th
e
d
etec
t
io
n
r
ate
o
f
h
ig
h
ly
s
im
ilar
f
o
r
g
ed
s
ig
n
at
u
r
es.
3.
4
.
Dis
cus
s
io
n
T
h
e
ex
p
er
im
e
n
tal
f
in
d
in
g
s
d
em
o
n
s
tr
ate
th
at
th
e
p
r
o
p
o
s
ed
S
S
-
ANFI
S
m
o
d
el
p
r
o
v
id
es
a
p
r
ac
tical
an
d
r
eliab
le
ap
p
r
o
ac
h
f
o
r
o
f
f
lin
e
s
ig
n
atu
r
e
v
er
i
f
icatio
n
in
ac
ad
e
m
ic
d
o
cu
m
en
t
a
u
th
en
ticatio
n
.
T
h
e
in
teg
r
atio
n
o
f
p
s
eu
d
o
-
lab
el
-
b
ased
s
elf
-
tr
ain
i
n
g
en
ab
les
th
e
m
o
d
el
to
u
t
ilize
u
n
lab
eled
s
am
p
les
d
u
r
i
n
g
tr
ain
in
g
.
T
h
is
m
ec
h
an
is
m
h
el
p
s
ex
p
an
d
th
e
ef
f
ec
tiv
e
tr
ain
in
g
s
et
with
o
u
t
r
eq
u
i
r
in
g
ad
d
itio
n
al
m
a
n
u
al
lab
elin
g
,
wh
ich
is
p
ar
ticu
lar
ly
r
elev
an
t
in
ac
ad
e
m
ic
en
v
ir
o
n
m
en
ts
wh
er
e
co
ll
ec
tin
g
lar
g
e
lab
eled
s
ig
n
at
u
r
e
d
atasets
is
d
if
f
icu
lt
an
d
tim
e
-
co
n
s
u
m
in
g
.
C
o
m
p
ar
ed
with
co
n
v
e
n
tio
n
al
ANFI
S,
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
b
en
ef
its
f
r
o
m
th
e
s
em
i
-
s
u
p
er
v
is
ed
r
ef
in
em
en
t
p
r
o
ce
s
s
.
C
o
n
v
e
n
tio
n
al
ANFI
S
r
elies
o
n
ly
o
n
lab
eled
s
am
p
les
to
co
n
s
tr
u
ct
f
u
zz
y
r
u
les
an
d
m
em
b
er
s
h
ip
f
u
n
ctio
n
s
.
I
n
co
n
tr
ast,
SS
-
ANFI
S
in
co
r
p
o
r
ates
h
ig
h
-
co
n
f
id
e
n
ce
p
s
eu
d
o
-
lab
el
ed
s
am
p
les
in
to
t
h
e
tr
ain
in
g
p
r
o
ce
s
s
,
allo
win
g
th
e
m
o
d
el
to
r
ef
in
e
its
d
ec
is
io
n
b
o
u
n
d
a
r
y
.
T
h
is
ex
p
lain
s
wh
y
th
e
p
r
o
p
o
s
ed
m
o
d
el
ac
h
iev
ed
h
i
g
h
er
p
r
ec
is
io
n
an
d
co
m
p
etitiv
e
o
v
e
r
all
class
if
icati
o
n
p
er
f
o
r
m
a
n
ce
.
T
h
e
h
ig
h
p
r
ec
is
io
n
ac
h
iev
ed
b
y
th
e
p
r
o
p
o
s
ed
m
o
d
el
is
o
n
e
o
f
th
e
m
o
s
t
im
p
o
r
tan
t
f
in
d
i
n
g
s
o
f
th
is
s
tu
d
y
.
I
n
p
r
ac
tical
ac
ad
em
ic
d
o
cu
m
en
t
v
er
i
f
icatio
n
,
f
alse
p
o
s
itiv
e
er
r
o
r
s
m
ay
ca
u
s
e
g
en
u
i
n
e
s
ig
n
atu
r
es
to
b
e
in
co
r
r
ec
tly
f
lag
g
e
d
as
f
o
r
g
e
d
.
Su
ch
e
r
r
o
r
s
ca
n
c
r
ea
te
u
n
n
e
ce
s
s
ar
y
ad
m
in
is
tr
ativ
e
p
r
o
b
le
m
s
.
T
h
er
ef
o
r
e,
th
e
ab
ilit
y
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el
to
m
in
im
ize
f
alse
p
o
s
itiv
e
p
r
e
d
ictio
n
s
m
a
k
es
it
s
u
ita
b
le
f
o
r
s
u
p
p
o
r
tin
g
v
er
if
icatio
n
task
s
in
ac
ad
em
ic
in
s
titu
tio
n
s
.
Nev
er
th
eless
,
th
e
lo
wer
r
ec
all
in
d
icate
s
th
at
s
o
m
e
f
o
r
g
e
d
s
i
g
n
atu
r
es
wer
e
s
till
class
if
ied
as
g
en
u
in
e
.
T
h
is
r
esu
lt
s
u
g
g
ests
th
at
h
ig
h
ly
s
im
ilar
f
o
r
g
ed
s
ig
n
at
u
r
es
r
em
ain
ch
allen
g
in
g
f
o
r
t
h
e
cu
r
r
e
n
t
f
ea
tu
r
e
r
ep
r
esen
tatio
n
.
Sin
ce
th
is
s
tu
d
y
u
s
es
o
f
f
lin
e
s
tatic
s
ig
n
atu
r
e
im
ag
es,
th
e
m
o
d
el
ca
n
n
o
t
ac
ce
s
s
d
y
n
am
ic
s
ig
n
in
g
in
f
o
r
m
atio
n
s
u
ch
as
wr
itin
g
s
p
ee
d
,
p
e
n
p
r
ess
u
r
e,
o
r
s
tr
o
k
e
o
r
d
er
.
As
a
r
esu
lt,
th
e
m
o
d
el
m
u
s
t
r
ely
en
tire
ly
o
n
s
tatic
v
is
u
al
p
atter
n
s
,
wh
ich
m
ay
n
o
t
alwa
y
s
b
e
s
u
f
f
icien
t
to
d
is
tin
g
u
is
h
s
k
illed
f
o
r
g
er
ies
f
r
o
m
g
en
u
in
e
s
ig
n
atu
r
es.
C
o
m
p
ar
ed
with
C
NN
-
b
ased
m
eth
o
d
s
,
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
o
f
f
er
s
a
m
o
r
e
in
ter
p
r
e
tab
le
an
d
lig
h
tweig
h
t
alter
n
ativ
e.
C
NN
m
o
d
els
ar
e
p
o
wer
f
u
l
f
o
r
im
ag
e
-
b
ased
r
ec
o
g
n
itio
n
task
s
,
b
u
t
th
ey
ty
p
ically
r
eq
u
ir
e
lar
g
er
lab
eled
d
atasets
an
d
h
ig
h
er
co
m
p
u
tatio
n
al
r
eso
u
r
ce
s
.
T
h
e
p
r
o
p
o
s
ed
SS
-
ANFI
S
m
o
d
el
is
m
o
r
e
s
u
itab
le
f
o
r
s
m
all
-
s
ca
le
in
s
tit
u
tio
n
al
ap
p
licatio
n
s
b
ec
au
s
e
i
t
co
m
b
in
es
f
u
zz
y
r
ea
s
o
n
in
g
,
c
o
m
p
ac
t
PC
A
-
b
ased
f
ea
tu
r
e
r
ep
r
esen
tatio
n
,
an
d
s
e
m
i
-
s
u
p
er
v
is
ed
lear
n
i
n
g
.
T
h
e
Stre
am
lit
-
b
ased
p
r
o
to
ty
p
e
also
d
e
m
o
n
s
tr
ates
th
e
p
r
ac
tic
al
ap
p
licab
ilit
y
o
f
th
e
p
r
o
p
o
s
e
d
m
eth
o
d
.
T
h
e
s
y
s
tem
allo
ws
u
s
er
s
to
u
p
lo
ad
s
ig
n
atu
r
e
im
ag
es
an
d
o
b
t
ain
p
r
ed
ictio
n
r
esu
lts
in
r
ea
l
tim
e.
T
h
is
p
r
o
to
ty
p
e
ca
n
s
u
p
p
o
r
t
ad
m
in
is
tr
ativ
e
s
t
af
f
in
s
cr
ee
n
in
g
s
u
s
p
icio
u
s
s
ig
n
atu
r
es
b
ef
o
r
e
f
u
r
th
er
m
an
u
al
v
er
if
icatio
n
is
co
n
d
u
cte
d
.
T
h
er
ef
o
r
e,
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
is
n
o
t o
n
ly
m
e
th
o
d
o
lo
g
ically
r
ele
v
an
t b
u
t
als
o
p
r
ac
tically
u
s
ef
u
l
f
o
r
ac
ad
e
m
ic
d
o
c
u
m
en
t a
u
t
h
e
n
ticatio
n
.
3.
5
.
L
im
it
a
t
io
ns
T
h
is
s
tu
d
y
h
as
s
ev
er
al
lim
it
atio
n
s
.
First,
th
e
d
ataset
was
co
llected
f
r
o
m
a
lim
ited
i
n
s
titu
tio
n
a
l
en
v
ir
o
n
m
en
t,
s
o
b
r
o
ad
er
ev
al
u
atio
n
ac
r
o
s
s
d
if
f
er
e
n
t
in
s
titu
tio
n
s
,
wr
iter
s
,
an
d
d
o
cu
m
en
t
ty
p
es
is
r
eq
u
ir
ed
.
Seco
n
d
,
th
e
s
tu
d
y
f
o
cu
s
es
o
n
l
y
o
n
o
f
f
lin
e
s
ig
n
at
u
r
e
im
ag
es;
th
er
ef
o
r
e
,
d
y
n
am
ic
s
ig
n
in
g
i
n
f
o
r
m
atio
n
s
u
ch
as
p
en
p
r
ess
u
r
e,
wr
itin
g
s
p
ee
d
,
an
d
s
tr
o
k
e
o
r
d
er
was
n
o
t
c
o
n
s
id
er
ed
.
T
h
ir
d
,
alth
o
u
g
h
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
ac
h
iev
ed
h
ig
h
p
r
ec
is
io
n
,
th
e
r
ec
all
v
alu
e
in
d
icate
s
th
at
s
o
m
e
f
o
r
g
ed
s
ig
n
atu
r
es
wer
e
s
t
ill
m
is
cla
s
s
if
ied
a
s
g
en
u
in
e.
T
h
is
lim
itatio
n
s
u
g
g
ests
th
at
m
o
r
e
d
is
cr
im
in
ativ
e
f
ea
tu
r
e
ex
tr
ac
tio
n
m
eth
o
d
s
ar
e
n
ee
d
ed
t
o
im
p
r
o
v
e
f
o
r
g
er
y
d
etec
tio
n
.
Fo
u
r
th
,
th
e
co
m
p
a
r
is
o
n
in
T
ab
le
4
is
p
r
esen
ted
as
a
liter
at
u
r
e
-
b
ased
co
n
te
x
tu
al
co
m
p
ar
i
s
o
n
r
ath
er
th
an
a
d
ir
ec
t
ex
p
er
im
en
tal
b
en
ch
m
ar
k
,
b
ec
a
u
s
e
th
e
r
ef
er
en
ce
d
m
eth
o
d
s
m
a
y
u
s
e
d
if
f
er
en
t
d
atasets
an
d
ev
alu
atio
n
p
r
o
to
co
ls
.
Fu
tu
r
e
wo
r
k
s
h
o
u
ld
im
p
lem
e
n
t
all
b
aselin
e
m
eth
o
d
s
o
n
th
e
s
am
e
d
ataset
an
d
u
n
d
er
th
e
s
am
e
ev
alu
atio
n
p
r
o
to
co
l
to
p
r
o
v
id
e
a
s
tr
icter
ex
p
er
im
en
t
al
co
m
p
a
r
is
o
n
.
Fu
r
th
er
s
tu
d
ie
s
m
ay
also
in
clu
d
e
R
OC
-
AU
C
an
aly
s
i
s
,
co
m
p
u
ta
tio
n
al
tim
e
ev
alu
atio
n
,
a
b
latio
n
s
tu
d
ies,
an
d
lar
g
e
r
-
s
ca
le
d
atasets
wi
th
s
k
illed
f
o
r
g
er
y
s
ce
n
ar
io
s
.
4.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
p
r
o
p
o
s
ed
a
Sem
i
-
Su
p
er
v
is
ed
Ad
ap
tiv
e
Neu
r
o
-
Fu
zz
y
I
n
f
er
en
ce
Sy
s
tem
(
SS
-
ANFI
S)
f
o
r
o
f
f
lin
e
s
ig
n
at
u
r
e
v
er
if
icatio
n
i
n
an
ac
ad
em
ic
d
o
c
u
m
en
t
a
u
th
en
ticatio
n
co
n
tex
t.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
in
teg
r
ates
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
4
,
Au
g
u
s
t
20
2
6
:
1
9
8
5
-
1
997
1994
p
s
eu
d
o
-
lab
el
-
b
ased
s
elf
-
tr
ain
i
n
g
in
t
o
th
e
ANFI
S
lear
n
in
g
p
r
o
ce
s
s
to
u
tili
ze
b
o
th
la
b
e
led
an
d
u
n
lab
eled
s
ig
n
atu
r
e
s
am
p
les.
E
x
p
er
im
e
n
tal
r
esu
lts
o
n
th
e
Dip
a
Un
iv
er
s
ity
Ma
k
ass
ar
d
ataset
s
h
o
w
th
at
th
e
p
r
o
p
o
s
ed
SS
-
ANFI
S a
ch
iev
ed
an
ac
cu
r
ac
y
o
f
9
0
.
5
%,
p
r
ec
is
io
n
o
f
9
8
.
8
%,
r
ec
all
o
f
8
2
.
0
%,
an
d
F1
-
s
co
r
e
o
f
9
0
.
0
%.
T
h
e
h
ig
h
p
r
ec
is
io
n
in
d
icate
s
th
at
th
e
m
o
d
el
is
ef
f
ec
tiv
e
in
m
in
im
izin
g
f
alse
p
o
s
itiv
e
p
r
ed
ictio
n
s
,
wh
ich
is
im
p
o
r
tan
t
f
o
r
s
ec
u
r
ity
-
s
en
s
itiv
e
ac
ad
em
ic
d
o
c
u
m
e
n
t
v
er
if
icatio
n
.
Ho
wev
er
,
th
is
s
tu
d
y
is
lim
ited
b
y
th
e
in
s
titu
tio
n
al
s
co
p
e
o
f
th
e
d
ataset
an
d
th
e
u
s
e
o
f
o
f
f
lin
e
s
tatic
s
ig
n
atu
r
e
im
ag
es
o
n
ly
.
Fu
tu
r
e
wo
r
k
will
f
o
cu
s
o
n
ex
p
an
d
i
n
g
th
e
d
atas
et
ac
r
o
s
s
m
o
r
e
u
s
er
s
,
in
co
r
p
o
r
atin
g
m
o
r
e
ch
allen
g
in
g
s
k
ille
d
f
o
r
g
e
r
y
s
ce
n
ar
io
s
,
an
d
ex
p
lo
r
in
g
d
ee
p
f
ea
tu
r
e
e
x
tr
ac
tio
n
o
r
h
y
b
r
id
o
p
tim
izatio
n
m
eth
o
d
s
to
im
p
r
o
v
e
r
ec
all
an
d
r
o
b
u
s
tn
ess
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
r
esear
ch
was
s
u
p
p
o
r
ted
b
y
Dip
a
Un
i
v
e
rsity
M
a
k
a
ss
a
r
th
r
o
u
g
h
th
e
I
n
ter
n
al
L
ec
tu
r
er
R
esear
ch
Sch
em
e,
u
n
d
er
Gr
an
t N
o
.
0
2
2
4
/UNDI
PA/G.4
/I
X/2
0
2
5
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
I
n
ac
co
r
d
an
ce
with
t
h
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
,
th
e
s
p
ec
if
ic
c
o
n
tr
ib
u
tio
n
s
o
f
ea
ch
au
th
o
r
a
r
e
p
r
o
v
id
ed
b
elo
w
to
e
n
s
u
r
e
tr
an
s
p
ar
e
n
cy
an
d
ac
co
u
n
tab
ilit
y
in
th
e
r
esear
c
h
an
d
p
u
b
licatio
n
p
r
o
ce
s
s
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Sad
ly
Sy
am
s
u
d
d
in
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
J
u
f
r
i
✓
✓
✓
✓
✓
✓
✓
Su
ci
R
ah
m
a
Dan
i
R
ac
h
m
an
✓
✓
✓
✓
✓
✓
✓
✓
✓
Su
ty
an
i
✓
✓
✓
✓
✓
W
ilem
Mu
s
u
✓
✓
✓
✓
✓
✓
✓
Salm
iati
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
T
h
e
au
t
h
o
r
s
d
ec
lar
e
th
at
th
e
y
h
av
e
n
o
k
n
o
wn
c
o
m
p
etin
g
f
in
an
cial
in
ter
ests
o
r
p
er
s
o
n
al
r
el
atio
n
s
h
ip
s
th
at
co
u
ld
h
av
e
ap
p
ea
r
ed
t
o
in
f
lu
en
ce
th
e
wo
r
k
r
e
p
o
r
te
d
in
t
h
is
p
ap
er
.
DATA AV
AI
L
AB
I
L
I
T
Y
All
d
ata,
as
well
as
th
e
p
r
o
g
r
a
m
s
o
r
ap
p
licatio
n
s
g
en
er
ate
d
d
u
r
in
g
th
is
s
tu
d
y
,
a
r
e
av
ailab
l
e
f
r
o
m
th
e
co
r
r
esp
o
n
d
in
g
a
u
th
o
r
u
p
o
n
r
e
q
u
est.
RE
F
E
R
E
NC
E
S
[
1
]
R
.
P
l
a
m
o
n
d
o
n
a
n
d
S
.
N
.
S
r
i
h
a
r
i
,
“
O
n
l
i
n
e
a
n
d
o
f
f
l
i
n
e
h
a
n
d
w
r
i
t
i
n
g
r
e
c
o
g
n
i
t
i
o
n
:
A
c
o
m
p
r
e
h
e
n
si
v
e
s
u
r
v
e
y
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
Pa
t
t
e
r
n
An
a
l
y
si
s
a
n
d
Ma
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
2
2
,
n
o
.
1
,
2
0
0
0
,
d
o
i
:
1
0
.
1
1
0
9
/
3
4
.
8
2
4
8
2
1
.
[
2
]
D
.
I
mp
e
d
o
v
o
a
n
d
G
.
P
i
r
l
o
,
“
A
u
t
o
m
a
t
i
c
s
i
g
n
a
t
u
r
e
v
e
r
i
f
i
c
a
t
i
o
n
:
Th
e
s
t
a
t
e
o
f
t
h
e
a
r
t
,
”
I
EE
E
T
r
a
n
sa
c
t
i
o
n
s
o
n
S
y
s
t
e
m
s,
M
a
n
,
a
n
d
C
y
b
e
r
n
e
t
i
c
s,
P
a
r
t
C
(
A
p
p
l
i
c
a
t
i
o
n
s
a
n
d
R
e
v
i
e
w
s)
,
v
o
l
.
3
8
,
n
o
.
5
,
p
p
.
6
0
9
–
6
3
5
,
2
0
0
8
,
d
o
i
:
1
0
.
1
1
0
9
/
TS
M
C
C
.
2
0
0
8
.
9
2
3
8
6
6
.
[
3
]
L.
G
.
H
a
f
e
m
a
n
n
,
R
.
S
a
b
o
u
r
i
n
,
a
n
d
L.
S
.
O
l
i
v
e
i
r
a
,
“
O
f
f
l
i
n
e
h
a
n
d
w
r
i
t
t
e
n
s
i
g
n
a
t
u
r
e
v
e
r
i
f
i
c
a
t
i
o
n
-
Li
t
e
r
a
t
u
r
e
r
e
v
i
e
w
,
”
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
7
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
I
m
a
g
e
Pr
o
c
e
ssi
n
g
T
h
e
o
r
y
,
T
o
o
l
s
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s,
I
PTA
2
0
1
7
,
v
o
l
.
2
0
1
8
-
Jan
u
a
,
p
p
.
1
–
8
,
2
0
1
7
,
d
o
i
:
1
0
.
1
1
0
9
/
I
P
TA
.
2
0
1
7
.
8
3
1
0
1
1
2
.
[
4
]
B
.
M
.
A
l
-
m
a
q
a
l
e
h
a
n
d
A
.
Q
.
M
.
M
o
h
a
mm
e
d
,
“
A
s
u
r
v
e
y
o
n
h
a
n
d
w
r
i
t
t
e
n
s
i
g
n
a
t
u
r
e
v
e
r
i
f
i
c
a
t
i
o
n
a
p
p
r
o
a
c
h
e
s,”
C
o
m
m
u
n
i
c
a
t
i
o
n
s
o
n
Ap
p
l
i
e
d
E
l
e
c
t
r
o
n
i
c
s
(
C
A
E)
,
v
o
l
.
4
,
n
o
.
8
,
p
p
.
2
3
–
2
9
,
2
0
1
6
.
[
5
]
M
.
A
.
F
e
r
r
e
r
,
J.
B
.
A
l
o
n
s
o
,
a
n
d
C
.
M
.
Tr
a
v
i
e
s
o
,
“
O
f
f
l
i
n
e
g
e
o
me
t
r
i
c
p
a
r
a
met
e
r
s
f
o
r
a
u
t
o
m
a
t
i
c
si
g
n
a
t
u
r
e
v
e
r
i
f
i
c
a
t
i
o
n
u
s
i
n
g
f
i
x
e
d
-
p
o
i
n
t
a
r
i
t
h
m
e
t
i
c
,
”
I
EE
E
T
r
a
n
s
a
c
t
i
o
n
s
o
n
P
a
t
t
e
r
n
A
n
a
l
y
si
s
a
n
d
M
a
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
2
0
0
5
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
0
5
.
1
2
5
.
[
6
]
J.
F
.
V
a
r
g
a
s,
M
.
A
.
F
e
r
r
e
r
,
C
.
M
.
Tr
a
v
i
e
s
o
,
a
n
d
J.
B
.
A
l
o
n
s
o
,
“
O
f
f
-
l
i
n
e
s
i
g
n
a
t
u
r
e
v
e
r
i
f
i
c
a
t
i
o
n
b
a
se
d
o
n
g
r
e
y
l
e
v
e
l
i
n
f
o
r
ma
t
i
o
n
u
si
n
g
t
e
x
t
u
r
e
f
e
a
t
u
r
e
s,
”
P
a
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
,
v
o
l
.
4
4
,
n
o
.
2
,
2
0
1
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
a
t
c
o
g
.
2
0
1
0
.
0
7
.
0
2
8
.
[
7
]
K
.
H
u
a
n
g
a
n
d
H
.
Y
a
n
,
“
O
f
f
-
l
i
n
e
s
i
g
n
a
t
u
r
e
v
e
r
i
f
i
c
a
t
i
o
n
u
s
i
n
g
s
t
r
u
c
t
u
r
a
l
f
e
a
t
u
r
e
c
o
r
r
e
s
p
o
n
d
e
n
c
e
,
”
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
,
v
o
l
.
3
5
,
n
o
.
1
1
,
p
p
.
2
4
6
7
–
2
4
7
7
,
2
0
0
2
,
d
o
i
:
1
0
.
1
0
1
6
/
S
0
0
3
1
-
3
2
0
3
(
0
1
)
0
0
2
2
2
-
9.
[
8
]
M
.
N
a
z
i
r
,
“
O
f
f
l
i
n
e
si
g
n
a
t
u
r
e
v
e
r
i
f
i
c
a
t
i
o
n
s
y
st
e
m:
a
n
o
v
e
l
t
e
c
h
n
i
q
u
e
o
f
f
u
si
o
n
o
f
G
LC
M
a
n
d
g
e
o
m
e
t
r
i
c
f
e
a
t
u
r
e
s
u
si
n
g
S
V
M
,
”
Evaluation Warning : The document was created with Spire.PDF for Python.